Académique Documents
Professionnel Documents
Culture Documents
0> (Douglass,
1971), where ND normal direction and TD transverse direction. Meanwhile later the
new detailed data were obtained concerning texture development in -Zr under rolling. In
particular, it was established that by rolling of a textureless slab the intermediate texture
(0001) 15-25 ND-RD <112
0}<101
0> (T3) are present often in textures of rolled sheets and, especially, tubes. All
these texture components form owing to activity of concrete combinations of plastic
deformation mechanisms, have their characteristic strain hardening and therefore show
Recrystallization 4
different tendencies to recrystallization. In the real case of a multicomponent texture the
development of recrystallization must be additionally complicated by interaction between
different components in regions of their contact. Indeed, according to (Isaenkova et al., 1988;
Perlovich et al., 1989), resulting changes of -Zr rolling textures in the course of
recrystallization can not be reduced to 30-rotation around basal normals and require for
more complex description. In order to investigate this question in more details, the
following work was undertaken.
2.1 Materials and methods
Recrystallization was investigated in sheet samples of alloys Zr-2,5%Nb, Zr-2,3%Cr and
pure Zr as well as in tube samples of the alloy Zr-2,5%Nb. Sheets were produced by
longitudinal or transverse cold rolling up to deformation degrees in the range from 40%
to 90% in such a way as to form the following textures: T1, T1+T2, T2, T1+T2+T3. The
weak component T0 was present everywhere. Perfection parameters and mutual
relationship of different components varied. The channel tube was cold-rolled by 50%-
thinning of its wall. All samples were annealed in dynamic vacuum at temperatures 500-
600C during 1-5 h.
The main used method was X-ray diffractometric texture analysis. Direct pole figures
(PF) (0001), {112
0} and {101
2}.
Main results concern reorientation of basal normals by -Zr recrystallization, i.e. changes of
PF(0001). Superposition of SD and PF(0001) is shown in Fig. 1 for the sheet alloy Zr-2,5%Nb
rolled up to deformation degrees 40, 60 and 80%, corresponding to formation of different
textures: T1, T1+T2, T2. The densest cross-hatching indicates zones, where texture are
localized predominantly. These zones are situated at slopes of initial texture maxima,
increasing scattering of the recrystallization texture. Hence, the model of inhomogeneous
strain hardening, proposed in (Perlovich, 1994) for textured BCC-metals, is true for HCP -
phase also. But while taking into account the multicomponent character of observed rolling
textures, regularities of recrystallization should be more complicated.
Development of Texture and Substructure
Inhomogeneity by Recrystallization of Rolled Zr-Based Alloys 5
Fig. 1. Superposition of partial PF(0001) (dotted contours) for rolled sheets of the alloy Zr-
2,5%Nb by deformation degrees 40% (a), 60% (b), 80% (c) and SD (solid contours),
characterizing texture changes by recrystallization. Cross-hatching density reflects an
increase of pole density by recrystallization.
For rolling textures T1+T2, where I(T1)>I(T2), it was established that the relative increase of
main different components in the recrystallization texture depends on the initial
relationship of these components in the rolling texture. In particular:
I(T2)/I(T1)=f
1
[I(T2)/I(T1)], (1)
I(T2)/I(T0)=f
2
[I(T2)/I(T0)]. (2)
These dependences are drawn in Fig. 2, where linearity of f
1
and nonlinearity of f
2
are
evident. The nearer are initial intensities of components T2 and T1, the greater is difference
between growth rates of new grains with corresponding orientations. As an area of the
Fig. 2. Relative changing of texture components T2 versus T1 (a) and T2 versus T0 (b) in the
recrystallization texture of sheet samples depending on the initial relationship of these
components in the rolling texture.
Recrystallization 6
contact surface between deformed grains of these components increases, conditions for
growth of grains with the orientation T2 become more favorable. This signifies that T2-
grains are growing into T1-grains and absorb them. The weaker dependence connects
growths of components T2 and T0 in the recrystallization texture, since the difference
between their strain hardening is less than in the case of components T2 and T1.
Recrystallization of samples, showing predominance of the component T1, involves
essential redistribution of basal normals even to the point of main texture component
changing; then, depending on the concrete relationship of components in the rolling texture,
T1 can give way to T0 (Fig. 3-a) or T2 (Fig. 3-b). Composition of the alloy influences mainly
temperature parameters of the recrystallization process in -Zr and seems to be of
secondary importance for orientation regularities.
Fig. 3. Partial PF (0001) for sheet alloys Zr-2,5%Nb (a) and Zr-2,3%Cr (b) in rolled (dotted
contours) and recrystallized (solid contours) states.
2.3 Recrystallization in tubes
The rolling texture of investigated tubes exhibits predominance of the component
{112
0}<101
0> and contains the intensive axial component with the axis <101
0} (b) for tubes of the alloy Zr-2,5%Nb: 1 initial rolled state; 2-6 annealing regimes:
2 - 500C, 3 h; 3 - 530C, 3 h; 4 - 550C, 3 h; 5 - 580C, 3 h; 5 - 580C, 1 h; 6 - 600C, 3h.
Recrystallization 8
As a result of tube annealing the new maxima arise on the meridional section of PF{112
0}
(Fig. 5-b), testifying that, along with redistribution of basal normals in investigated tubes,
reorientation of prismatic normals takes place also, leading to development of the typical for
-Zr recrystallization texture (0001) R-T <213
0 112
0} for surface,
intermediate and central layers of the hot-rolled sheet are presented in Fig. 9. When passing
from the surface layer of sheet to the central one, the rolling texture changes essentially. The
following characteristic features of the rolling texture are considered: (1) the angular distance
of texture maxima from normal direction (ND) in PF(0001); (2) the presence of additional
texture maxima in PF{112
0} 70.
0 800 1600
t, m
10
30
50
70
0 800 1600
t, m
0.0
0.8
1.6
,
deg.
P
recr
P
def
a
b
A
B
A
B
Fig. 10. Layer-by-layer changes of texture characteristics across the thickness of hot-rolled
sheets: a angular distance of texture maxima from ND in PF(0001); b - ratio P
recr
/P
def
,
measured by PF{112
0}.
TD
RD
1
2
3
5
7
10
RD RD
RD RD RD
D R
(0001)
{112
0}
Development of Texture and Substructure
Inhomogeneity by Recrystallization of Rolled Zr-Based Alloys 15
4.2 Dynamic recrystallization under rolling
Development of recrystallization in -Zr under rolling is seen in the surface layer up to 0.8
mm thick (Fig. 10-b). In PF{112
{011}
, <112
0>
<111>
.
Multiplication of initial orientations in consequence of PT was analyzed in
(Cheadle & Ells, 1966). By absence of variant selection, 35 new orientations of the basal plane
arise in addition to the initial one; by mutual coincidence of some orientations their total
number decreases down to 24. Since in reality we deal with textured polycrystals of Zr-
alloys instead of single crystals, after PT the resulting distribution of basal axes in
PF(0001) consists of overlapping texture maxima rather than of separate points. Therefore
some new orientations of 24 above-mentioned ones, being close to each other, form common
maxima. Thus, PT complicates an initial texture, multiplying its maxima in a definite way,
though the resulting PF(0001) of the treated sample contains a lesser number of separate
maxima, than the Burgers relationship predicts.
5.2 Studied samples and investigation technique
Samples for investigation were cut from sheets of the Zr-2,5%Nb alloy, cold-rolled up to the
deformation degree of ~80%. In order to prepare semi-recrystallized samples, after cold
rolling they were annealed at 550C for 1 h. In order to induce PT , both cold-rolled
and annealed samples were subjected to the heat treatment in dynamic vacuum, including
heating up to 950C for 0,25 h, holding at this temperature during 0,5 h and subsequent
cooling with an evacuated envelope in air. X-ray texture measurements were carried out by
the standard method (Borodkina & Spector, 1981) using the diffractometer DRON-3M and
Cu K
0}<112
0> arises, having its maximum in PF(0001) at TD. Other maxima of the same set
are present at the predicted regions also, confirming additionally the fact of preliminary
recrystallization in the course of heating to the -phase. At the same time, maxima,
originating from deformed -grains, by PT become noticeably weaker, than in the case of PT
without preliminary recrystallization (compare Fig. 11-b and 12-b). The real situation in the
deformed textured -phase requires a description, similar to the case of a composite;
therefore it would be correct to state a possibility of different outcomes from the competition
between recrystallization and PT depending on deformation degree, grain orientation,
heating rate, and so on. Inhomogeneous development of the considered processes, as well as
their mutual competition, corresponds apparently to the most general case.
5.4 Recrystallization in the thermal influence zone by welding
Below some observations are presented concerning the competition between
recrystallization and PT in the thermal influence zone (TIZ) with reference to the Zr-
2,5%Nb alloy.
Arc welding is accompanied by a local heat treatment of the material in the vicinity of the
welding seam. Parameters of a short-time thermal cycle, passing lengthwise TIZ parallel to
the welding direction, are different for each longitudinal section of TIZ and depend on its
distance from the central line of the seam. Layer-by-layer study of the texture within the TIZ
gives an insight into the inhomogeneous structure developed in this zone by welding. In
Fig. 13-a a schematic image of a welded joint is drawn; the melting zone is denoted by dense
hatching and TIZ by thin hatching. Three longitudinal sections are shown within the TIZ,
and for each section an arrow indicates the corresponding PF(0001), obtained by X-ray
diffractometric study just of this section. RD is brought into the centre of these PF in contrast
with above-presented PF(0001) in Fig. 11 and 12.
Judging from PF in Fig. 13, different textures have formed in the shown sections of the TIZ
depending on the distance from the seam. While in the most remote section the initial
distribution of basal axes remains unchanged, the textures of the two closer sections contain
new components produced by PT. Pole figures for these sections differ in relationship of PT
components, originated from deformed (A) and recrystallized (B) -grains. A quantitative
treatment of obtained experimental data included the calculation of parameters,
characterizing the relative contributions of both deformed and recrystallized components in
the measured texture. In Fig. 13-b the results of such treatment for 18 successive sections of
the TIZ are presented. The upper curve, constructed by PF{112
b
/ d
(1)
where:
F
V
is the volume fraction of the particles
d is the particle size
b
is the boundary energy
If particles are randomly distributed, the pinning pressure will be directionally isotropic. On
the other hand, if the distribution of the particles is anisotropic, there will be an anisotropic
pinning pressure on the boundaries. Figure 5 shows that the particles in the as-received
strips were aligned along the rolling direction. The pinning pressure parallel to the rolling
plane should be lower than that along the thickness direction. Therefore, the plate-like
recrystallized grain shape can be mainly attributed to the planar distribution of the particles.
The directional distribution of the particles might be driven by the rolling of the strip. A
plate-like morphology of the recrystallized grains is often reported in the recrystallization
behaviour of other dispersion-strengthened alloys (Klug et al., 1996; Chou, 1997), although
other dispersion-strengthened alloys show equiaxed recrystallized grain structures
(Miodownik et al., 1994; Miodownik et al., 1995).
Recrystallization of Dispersion-Strengthened Copper Alloys
37
Another unique recrystallization characteristic of the alumina DS copper alloy is that the
recrystallized grains are very large. Early researchers (Singer & Gessinger, 1982; Mino et al.,
1987; Kusunoki et al., 1990) reported that the very large recrystallized grains found in
dispersion-strengthened alloys are formed through secondary recrystallization. They
concluded that primary recrystallization occurred immediately before secondary
recrystallization, or during plastic deformation dynamic recrystallization. Later studies
(Klug et al., 1996) suggested that primary recrystallization was responsible for the formation
of large grains because microstructural changes are driven by stored energy acquired from
plastic deformation. While plastically deformed alumina DS copper alloy possesses a
sufficient driving force for recrystallization, a barrier to recrystallization exists due to the
particle pinning effect. Microstructural inhomogeneity, such as large bands, provides
preferential nucleation sites, and a large nucleus at a large band can grow with a size
advantage over the surrounding matrix. Therefore, the emergence of very large
recrystallized grains is a result of preferential nucleation at pre-existing large bands. The
annealing behaviour of alumina DS copper alloy might be regarded as secondary
recrystallization since very large recrystallized grains are formed when they overcome the
particle-pinning pressure. However, the microstructure of the alumina DS copper alloy
suggests that the driving force for recrystallization is stored energy by plastic deformation.
Thus, while the evolution of the annealed alumina DS copper alloy microstructure appears
to be due to secondary recrystallization, the mechanism that forms the very large
recrystallized grains is due to primary recrystallization.
4.2 Unique recrystallization texture
The recrystallization texture of the annealed alumina DS copper alloy can be
approximated by {112}<312>. To our knowledge, this texture has not been reported for
other copper alloys. The recrystallization texture is determined by the orientations of the
new grains and their growth rates. The present study discussed the role of these two
factors and how they determine the unique recrystallization texture of the alumina DS
copper alloy.
4.2.1 Selective nucleation
As discussed previously, pre-existing large bands provided favourable nucleation sites for
recrystallization. Pre-existing large particles could introduce particle deformation zones
that act as nucleation sites. Figure 22 shows the recrystallizing grains formed around the
particles and their orientations observed in the as-received strips after rapid annealing.
The orientation of grain A was similar to that of the deformed matrix, and multiple
twinning could cause grains B and C to generate different orientations. It is known that
PSN usually gives rise to weak recrystallization textures (Humphreys & Hatherly, 1995).
Band coalescence is unlikely but possible when the pinning of the boundary movement is
relaxed. Figure 23 shows that the band growth took place by coalescence of similarly
oriented bands. Various orientations could be generated from new grains resembling the
matrix orientations through PSN and band coalescence, as well as by subsequent
twinning. Since no specific grain orientations dominated as the new grains evolved, the
well-developed strong recrystallization texture {112}<312> could not be caused by new
grain evolution.
Recrystallization
38
4.2.2 Selective growth
According to the theory of selective growth, the recrystallization texture is determined by
the relative growth rates of the boundaries. The velocity of the moving boundary (V) is a
function of the boundary mobility (M) and the driving pressure (P), given by:
V=MP (2)
P can be expressed as follows:
Fig. 22. Longitudinal section TEM micrograph showing individual grains around a particle
and their orientations in the as-received strips annealed at 923 K for 10 s
Fig. 23. Longitudinal section TEM micrograph showing individual grains and their
orientations in the cold-rolled strips annealed at 1123 K for 3 s (Kim & Lee, 2002)
Recrystallization of Dispersion-Strengthened Copper Alloys
39
P=P
D
- P
C
= P
D
- 2
b
/R (3)
where:
P
D
is the stored energy,
P
C
is the opposing pressure from the boundary curvature
b
is the boundary energy
R is the radius of the grain.
In particle-strengthened alloys, the Zener pinning pressure (P
Z
) arises from the particles,
and P can be expressed as follows (Humphreys & Hatherly, 1995):
P=P
D
- P
C
P
Z
= P
D
- 2
b
/R 3F
V
b
/d (4)
where:
F
V
is the volume fraction of the particles
d is the particle size.
Recrystallizing grains will grow only when P is positive. P increases with increasing grain
size and decreasing boundary energy. The low-angle boundaries and twin boundaries have
a lower boundary energy than the high-angle boundaries. Based on Equation 4, only large
grains with low-angle boundaries or twin boundaries can overcome the pinning pressure.
High-angle boundaries can be stagnant, even though they have higher mobility than low-
angle boundaries. The recrystallization texture {112}<312> is defined as ND-rotated copper,
which is occasionally found in large bands in the deformed state, as shown in Figures 19
and 21. Recrystallizing grains with {112}<312> orientations have a chance to face the
surrounding deformed matrix with low-angle boundaries because {112}<312> orientations
deviate slightly from the deformation texture. Furthermore, {112}<312> orientations have a
twinning relationship between the two equivalent orientations among them. Figure 24
shows the orientations of two adjacent recrystallized grains observed in the cold-rolled and
annealed strips. The boundary shape and orientation relationship indicated that the grain
boundary of the two adjacent recrystallized grains was a twin boundary. Therefore, the
unique recrystallization texture was determined by the preferential growth of large
recrystallizing grains with low-angle boundaries or twin boundaries, even though those
boundaries had low mobility.
Fig. 24. Longitudinal section TEM micrograph and (111) pole figure showing two adjacent
recrystallized grains of the cold-rolled strips annealed at 1123 K for 1 hr
Recrystallization
40
4.3 Dependency of recrystallization on prior rolling conditions
As described in Section 3.2, the response to annealing of the alumina DS copper alloy is
influenced by prior rolling conditions. The annealing behaviour of the cold-rolled strip is
characterized by recrystallization, whereas recovery by band growth occurs in the hot-rolled
strip. Similar results have been reported for other dispersion-strengthened alloys (Petrovic
& Ebert, 1972; Singer & Gessinger, 1982). As recovery and recrystallization are competitive
processes, dynamic recovery during hot rolling could reduce the potential energy in the
alloy. This argument would also apply when comparing results between the cold-rolled and
hot-rolled strips; since the hardness of the hot-rolled strip is lower, the recovery process
during hot-rolling is governed by normal band growth. In our tests, after annealing, a
continuous band growth occurred in the hot-rolled strip, which became a coarse band
structure with high-angle boundary characteristics (see Table 7).
Subjecting the as-received strips to hot rolling gave rise to band growth and increased
the hardness. Plastic deformation during hot rolling could increase the dislocation
density, increasing the hardness. Therefore, a reduction in potential energy may not
occur during hot rolling. It is not clear at this time why the hot-rolled strips became
resistant to recrystallization. One explanation would be the homogeneity of the
microstructural evolution. Microstructural inhomogeneity often occurs during plastic
deformation, and these regions are frequently sites of initial recrystallization. The
deformation becomes more homogeneous as the deformation temperature
increases (Humphreys & Hatherly, 1995). A reduction in microstructural inhomogeneity
during hot rolling could be responsible for the suppression of discontinuous
recrystallization.
An alternative explanation is based on the assumption that coarse particles are sheared into
finer particles during hot rolling (Kim & Lee, 2002). The shear strength of the particle might
decrease with increasing temperature. Particle shearing could result in a decrease in
interparticle spacing, which in turn could give rise to the higher hardness and the
corresponding difficulty in recrystallization.
5. Conclusions
The recrystallization behaviour of boron-added alumina DS copper alloy strips was studied.
The results may be summarized as follows.
Recrystallization occurred only in the centre region of the strips. Pre-existing large bands
provided a favourable nucleation site for very large recrystallized grains.
The morphology of the recrystallized grains was plate-like due to the planar alignment of
the dispersed particles.
The recrystallization texture was indexed to {112}<312>. Preferential growth of the large
recrystallizing grains against the particle pinning appeared to determine this unique
recrystallization texture.
The hot-rolled strip underwent recovery accompanied by continuous band growth, but
without recrystallization.
Recrystallization of Dispersion-Strengthened Copper Alloys
41
6. References
Afshar, A. & Simchi, A. (2008). Abnormal Grain Growth in Alumina Dispersion-
Strengthened Copper Produced by An Internal Oxidation Process. Scripta
Materialia, Vol.58, No.11, (June 2008), pp.966-969, ISSN 1359-6462
Chou, T.S. (1997). Recrystallization Behavior and Grain Structure in Mechanically Alloyed
Oxide Dispersion Strengthened MA956 Steel. Materials Science and Engineering A,
Vol.223, No.1-2, (February 1997), pp. 78-90, ISSN 0921-5093
Gallagher, D.E. ; Hoyt, E.W. & Kirby, R.E. (1988). Surface Segregation of Boron in
Dispersion-Strengthened Copper. Journal of Materials Science, Vol.27, No.21
(November 1992), pp. 5926-5930, ISSN 0022-2461
Gessinger, G.H. (1976). Mechanical Alloying of IN-738. Metallurgial Transactions A, Vol.7,
No.8, (August 1976), pp. 1203-1209, ISSN 0360-2133
Hirsch, J. & Lcke, K. (1988). Mechanism of Deformation and Development of Rolling
Textures in Polycrystalline FCC Metals-I. Description of Rolling Texture
Development in Homogeneous CuZn Alloys. Acta Metallurgica, Vol.36, No.11,
(November 1988), pp. 2863-2882, ISSN 1359-6454
Humphreys, F.J. & Hatherly, M. (1995). Recrystallization and Related Annealing Phenomena,
Pergamon, ISBN 0-08-041884-8, Oxford, United Kingdom
Kim, S.-H. & Lee, D.N. (2001). Recrystallization of Alumina Dispersion Strengthened
Copper Strips. Materials Science and Engineering A, Vol.313, No.1-2, (August 2001),
pp. 24-33, ISSN 0921-5093
Kim, S.-H. & Lee, D.N. (2002). Annealing Behavior of Alumina Dispersion-Strengthened
Copper Strips Rolled Under Different Conditions. Metallurgical and Materials
Transactions A, Vol.33, No.6 (June 2002), pp. 1605-1616, ISSN 1073-5623
Klug, R.C. ; Krauss, G. & Matlock, D.K. (1996). Recrystallization in Oxide-Dispersion
Strengthened Mechanically Alloyed Sheet Steel. Metallurgical and Materials
Transactions A, Vol.27, No.7, (July 1996), pp. 1945-1960, ISSN 1073-5623
Kusunoki, K. ; Sumino, K. ; Kawasaki, Y. & Yamazaki, M. (1990). Effects of the Amount of
and Oxide Content on the Secondary Recrystallization Temperature of Nickel-Base
Superalloys. Metallurgical Transations A, Vol.21, No.2, (February 1990), pp.547-555,
ISSN 0360-2133
Matthies, S. ; Vinel, G.W. & Helming, K. (1987). Standard Distribution in Texture Analysis,
Akademie-Verlag, ISBN 3-05-500249-0, Berlin, Germany
Mino, K. ; Nakagawa, Y.G. & Ohtomo, A. (1987). Abnormal Grain Growth Behavior of An
Oxide Dispersion Strengthened Superalloy. Metallurgical Transactions A, Vol.18,
No.6, (June 1987), pp. 777-784, ISSN 0360-2133
Miodownik, M.A. ; Martin. J.W. & Little. E.A. (1994). Secondary Recrystallization of Two
Oxide Dispersion Strengthened Ferritic Superalloys : MA956 and MA 957. Materials
Science and Technology, Vol.10, No.2, (February 1994), pp. 102-109, ISSN 0267-0836
Miodownik, M.A. ; Humphreys, A.O. & Martin, J.W. (1995). Growth of Secondary
Recrystallized Grains during Zone Annealing of Oxide Dispersion Strengthened
Alloys. Materials Science and Technology, Vol.11., No.5, (May 1995), pp. 450-454, ISSN
0267-0836
Nadkarni, A. (1984). Dispersion Strengthened Copper Properties and Applications, In : High
Conductivity Copper and Aluminum Alloys, E. Ling, P.W. Taubenblat, (Ed.), 77-101,
TMS-ASME, Warrendale, PA
Recrystallization
42
Petrovic, J.J. & Ebert, L.J. (1972). Electron Microscopy Examination of Primary
Recrystallization in TD-Nickel. Metallurgical Transactions, Vol.3, No.5, (May 1972),
pp.1123-1129, ISSN 0360-2133
Petrovic, J.J. & Ebert, L.J. (1972). Abnormal Grain Growth in TD-Nickel. Metallurgical
Transactions, Vol.3, No.5, (May 1972), pp.1131-1136, ISSN 0360-2133
Preston, O. & Grant, N.J. (1961). Dispersion Strengthening of Copper by Internal Oxidation.
Transactions of the Metallurgical Society of AIME, Vol.221, (February 1961), pp.164-173
Randle, V. (1993). The Measurement of Grain Boundary Geometry, Institute of Physics
Publishing, ISBN 0-7503-0235-6, London, United Kingdom
Simchi, H. & Simchi, A. (2009). Tensile and Fatigue Fracture of Nanometric Alumina
Reinforced Copper with Bimodal Grain Size Distribution. Materials Science and
Engineering A, Vol.507, No.1-2, (May 2009), pp. 200-206, ISSN 0921-5093
Singer, R.F. & Gessinger, G.H. (1982). The Influence of Hot Working on the Subsequent
Recyrstallization of a Dispersion Strengthened Superalloy-MA 6000. Metallurgical
Transactions A, Vol.13, No.8, (August 1982), pp. 1463-1470, ISSN 0360-2133
Stephens, J.J. & Nix, W.D. (1985). The Effect of Grain Morphology on Longitudinal Creep
Properties of INCONEL MA 754 at Elevated Temperature. Metallurgical
Transactions A, Vol.16, No.7, (July 1985), pp. 1307-1324, ISSN 0360-2133
Sumino, Y. ; Watanabe, H. & Yoshida, N. (2009). The Microstructural Evolution of
Precipitate Strengthened Copper Alloys by Varing Temperature Irradiation. Journal
of Nuclear Materials, Vol.386-388, No.C, (April 2009), pp. 654-657, ISSN 0022-3115
Young, C.T. ; Steele, J.H. & Lytton, J.L. (1973). Characterization of Bicrystals Using Kikuchi
Patterns, Machine Design, Vol.4, No.9, (September 1973), pp. 2081-2089, ISSN 0024-
9114
3
Application of Orientation Mapping in TEM and
SEM for Study of Microstructural Evolution
During Annealing Example: Aluminum Alloy
with Bimodal Particle Distribution
K. Sztwiertnia, M. Bieda and A. Kornewa
Polish Academy of Sciences, Institute of Metallurgy and Materials Science, Krakow,
Poland
1. Introduction
There are still considerable gaps in the understanding of the recrystallization processes of
metallic materials, which reduce the possibility of controlling their course and introducing
technological modifications aimed at obtaining desirable properties. The lack of a complete
explanation can be attributed to the high complexity of the phenomenon, which consists of a
superposition of the local nucleation and grain growth processes. These processes depend
strongly on the characteristics of the matrix, which is typically complex and
heterogeneously deformed. The phenomenology of the process and its energetic causes are
known because they were examined long ago, e.g., (Humphreys & Hatherly, 2002). On the
other hand, the relevant physical mechanisms that control the nucleation and growth of new
grains are not entirely clear. This uncertainty exists, among other reasons, because the origin
of the crystallographic orientations of the nuclei is usually not known.
1.1 Crystallographic orientation and orientation characteristics of materials
The crystallographic orientation is a feature of a material that is defined at any point of the
sample at which the ordering of the crystal lattice is not disturbed (or not significantly
disturbed). It can be generally said that almost all of the basic quantities that characterize a
polycrystalline material and its properties have a direct or complex relationship to the
orientation (g
i
), which is a function of the coordinates x
i
, y
i
, z
i
of a point in the sample. The
g(x,y) function described in a plane of the sample defines the orientation topography
(commonly called the orientation map). According to the definition of the term, the
orientation at any point (x
i
, y
i
) in the sample is given by a rotation that brings the local
sample reference system with its origin at the point (x
i
, y
i
) into coincidence with the crystal
reference system. The orientation is described unambiguously by three parameters, which
can be expressed in different ways. Usually, for the convenience of calculation, the Euler
angles
1
, ,
2
are applied. In some cases, the parameters of the rotation axis (, ) and the
rotation angle are used because this scheme is easy to visualize. If the orientation is
described by a greater number of parameters, then the parameters depend on each other.
Recrystallization
44
Such is the situation in the case of crystallographic indices {hkl}<uvw>, commonly used in
practice. Extensive analysis of orientation problems can be found in (Morawiec, 2004). If an
orientation map is obtained, the grains or subgrains may be reconstructed by identifying areas
whose pixels have orientations within a specified range. The knowledge of the orientation
topography enables the identification of grain and subgrain boundaries, as well as other
microstructural inhomogeneities, by selection of misorientations between neighboring
measuring points. This approach enables stereological analysis with regard to the
crystallographic orientation. Based on the orientation topography, the orientation characteristics
of the microstructure can be determined (Pospiech et al., 1993). The set of orientation
characteristics comprises the "principal distributions", which are texture functions determined
by the whole set of measurements and "partial distributions", in which only part of
measurements are needed. The most important of the principal distributions is the well-known
orientation distribution function (ODF), which describes the crystallographic texture of a
material. The ODF is defined by the density of the global orientation distribution of the grains
(taking into account their volume fraction). Another "principal distribution is the orientation
difference distribution function (ODDF). The ODDF contains all possible misorientations
between the measured orientations. To investigate local textures (or microtextures) in selected
areas of inhomogeneities or in the environment of preferred orientations, partial distribution
functions are applied. The most important of these functions is the misorientation distribution
function (MODF), which describes the distribution of misorientation between the nearest
neighbor grains. The other partial orientation distributions are statistical quantities related to
orientation and misorientation, which may be related to the properties of a material, its
anisotropy or specific stages through which material passes (Pospiech et al., 1993).
Many of the essential properties of polycrystalline materials and their anisotropy depend
directly or indirectly on the topographical arrangement of orientations. Using the
orientation topography, the material properties that depend on the character and
distribution of the grain boundaries can be described, for example, segregation or corrosion.
The knowledge of the orientation topography is of basic importance for the understanding
of many processes that occur in the material, such as deformation, recrystallization, phase
transformation or diffusion.
It is, therefore, not surprising that the characterization of the microstructure based on the
sets of measured orientations has advanced as a well-established technique, known as
Orientation Microscopy (OM). The main concept behind this technique is the automatic
collection and indexing of many electron diffraction patterns that are correlated with sample
coordinates. The development of new generations of computer-controlled electron
microscopes has improved their spatial resolution and increased the rate at which large sets
of Electron Back Scattered Diffraction (EBSD) patterns can be collected and processed in a
Scanning Electron Microscope (SEM), e.g., (Dingley, 1984; Wright & Adams, 1992; Adams &
Dingley, 1994; Schwartz et al., 2009). Systems created orientation topographies using EBSD
in SEM are now very common, and commercially available versions of this technology are
essentially fully automated, e.g., (HKL, 2007; TSL, 2007).
1.2 Orientation imaging microscopy for recrystallization study
Obviously, the analysis and modeling of the recrystallization of a deformed metallic
material requires description of the microstructure evolution during annealing that is as
Application of Orientation Mapping in TEM and SEM for Study of Microstructural
Evolution During Annealing Example: Aluminum Alloy with Bimodal Particle Distribution
45
complete as possible. Such a description may be based on the orientation topographies
obtained by OM techniques in systematic measurements of a sample that undergoes a
specific deformation and annealing process, e.g., (Zaefferer et al., 2001; Sztwiertnia, 2008).
To study the recrystallization (particularly its early stages), a high spatial resolution in the
orientation measurement is required. Unfortunately, the spatial resolution that can be
achieved by EBSD/SEM measurements is relatively low. It falls approximately one order of
magnitude behind the spatial resolution in conventional SEM imaging, and still further
behind when compared to the spatial resolution of a transmission electron microscope
(TEM). The measurement is limited in this way because the inherent resolution of EBSD is
governed not by the diameter of the beam spot at the point of impact on the surface, but
primarily by the excitation volume. This quantity is the fraction of the interaction volume of
the primary electrons within the sample from which the pattern-forming electrons are back
diffracted and leave the crystal, without further scattering. This volume is strongly
dependent on the type of electron gun and the material being investigated. Tilting of the
sample during the EBSD measurement further degrades the spatial resolution and produces
resolution along the beam direction on the sample surface that is approximately three times
worse than the resolution along the direction perpendicular to the beam, e.g., (Schwartz et
al., 2009). For these reasons, the best achievable spatial resolution in EBSD/SEM special
cases is in the order of approximately 30 nm; however, the practical limit is approximately
100 nm. This limitation restricts the utility of EBSD/SEM for the investigation of very fine-
grained and deformed microstructures, as in the case of the early stages of recrystallization.
To obtain better spatial and angular resolution, similar systems have been developed for
TEM, e.g., (Haessner et al., 1983; Dingley, 2006; Morawiec, 1999; Morawiec et al., 2002;
Rauch & Dupuy, 2005). Despite some restrictions, such as the currently unsolved problems
of image analysis, difficulties in measurement automation and sample preparation, the OIM
technique applied to TEM offers spatial resolution better than 10 nm and can be used for
quantitative analysis of structures at the nanoscale. Such a system, built at the Institute of
Metallurgy and Materials Science (Morawiec et al., 2002; Sztwiertnia et al., 2006; Bieda-
Niemiec, 2007), was used for a study of the recrystallization of 6013 aluminum alloy
(Sztwiertnia et al., 2007; Bieda et al., 2010).
The 6013 aluminum alloy was chosen as a prototype material that represents a group of
commercial alloys with a bimodal second phase particle distribution. The second-phase
particles are used to control the strengthening, grain size and texture of the alloy. Such
alloys can be interesting for examination of the role of the second phase particles in the
recrystallization process, e.g., (Humphreys & Hatherly, 2002; Sztwiertnia et al., 2005;
Ardakani & Humphreys, 1994).
To elucidate the mechanisms of the alloy microstructure transformation during annealing, in
situ TEM experiments and combined calorimetricmicroscopic investigations of bulk
samples were carried out. The in situ experiments were necessary to provide information
about the temporal relationships between changes that occur in the metal at the beginning of
recrystallization. Dynamic studies using SEM or TEM should be capable of providing the
required information. Because of its already mentioned limitations, the EBSD/SEM
measurements of localized strain in the deformed polycrystal, which are particularly
interesting as nucleation sites, give rather poor information about the orientation. For the in
situ studies of such regions, measurements using convergent beam electron diffraction
Recrystallization
46
(CBED) or microdiffraction in a TEM are more suitable, although the proximity of the free
surface in the thin foils can be a complicating factor during annealing. The first in-situ TEM
observations, obtained by Bailey in 1960 and Hu in 1963, indicated differences in the
recrystallization processes that can occur during the annealing of bulk samples and thin
foils. As a consequence, many researchers have been skeptical about the results of such
experiments up to now. However, the results obtained later by other authors e.g., (Roberts &
Lehtinen, 1972; Hutchinson & Ray, 1973; Sztwiertnia & Haessner, 1994) allow the
determination of the experimental conditions, which ensures that the changes directly
observed in an annealed foil are at least similar to those occurring in a bulk sample. In
general, recrystallization is easiest in orthogonal sections from a rolled sheet in which the
grain boundaries, extending from top to bottom of the foil, present the most favorable
distribution of driving potential for migration. Grooving grains are not strongly inhibited in
the thin foil regions and frequently extend nearly to the edge of the foil (Hutchinson & Ray,
1973). Nevertheless, because of the thermal grooving, the recrystallization front always
stops in foil thinner than a certain critical value, which approximately depends on the
fineness of its microstructure. One can increase the usable foil thickness by increasing the
accelerating voltage. The impact of the sample thickness on grain boundary movement
explained in greater detail by Roberts & Lethinen, 1972.
In highly deformed 6013 aluminum alloy, the critical thickness for foils cut from planes
perpendicular to the sheet is so low that it allows in situ observation of nucleation events
(and to some extend the growth of the nuclei) to be carried out in a conventional TEM
operated at 200 kV (Sztwiertnia et al., 2005; Sztwiertnia et al., 2007; Bieda et al., 2010). To
examine the significance of the in situ experiments, the thin foils annealed in the TEM were
compared to thin foils prepared from bulk samples heated in a calorimeter to obtain a
specified recrystallization stage. The comparison shows that the processes occurring in both
types of foils were at least qualitatively the same.
2. Example: Recrystallization of aluminum alloy with bimodal particle
distribution
2.1 Material and investigation methodology
The changes of the microstructure during annealing were examined in the case of the
polycrystalline aluminum alloy 6013 (Table 1), which was previously identified as the
prototype of materials with a bimodal precipitate distribution.
Mg Si Cu Mn Fe others Al
1.15 1.0 1.1 0.3 0.5 0.15 remainder
Table 1. 6013 aluminum alloy chemical composition (% wt.).
Samples for testing were supersaturated, then aged and reversibly cold-rolled up to the 75 and
90 % of the 10 mm value. The deformed samples were examined by means of non-isothermal
annealing in a differential calorimeter (DC). It was found that the spectrum of released stored
energy contained several peaks. Next, a new series of samples were heated in the calorimeter
to the selected temperatures, rapidly cooled, and then analyzed in the TEM. The tests in the
TEM were complemented by SEM examinations. Because of the poor quality of the orientation
Application of Orientation Mapping in TEM and SEM for Study of Microstructural
Evolution During Annealing Example: Aluminum Alloy with Bimodal Particle Distribution
47
topographies measured by the standard EBSD/FEG/SEM techniques, these measurements
were used only to detect the presence of newly recrystallized grains (that is, only those
diffraction patterns were taken into account that had a high image quality and showed local
areas with low dislocation densities). The high degree of deformation also resulted in low
quality TEM diffraction patterns. However, it was still possible to measure enough single
orientations in the TEM to construct orientation topographies for all of the microstructural
elements of the cold-rolled material. The investigations also included a comparison of the
deformed alloy microstructure with that of the pure metal and the investigation of the
recrystallization process dynamic. The latter investigation consisted of the in situ
measurements in the TEM. This measurement was necessary to obtain information about the
time sequence of changes occurring in the material at the beginning of the recrystallization. In
the investigated material, the sequence of events that occurs in the deformation zones is of
particular interest because such areas undergo intense nucleation.
2.2 Deformation state
First, the deformation microstructure of the alloy was compared with that of the pure metal.
The microstructure of commercially available pure aluminum (3N Al), reversibly cold rolled
to 90%, has been chosen for comparison. Both microstructures are built of elongated in the
rolling direction (RD) and lie nearly parallel to the sheet plane grains and subgrains, as
shown in Fig. 1 b and 2 a. Some significant differences became evident when the measured
orientation topographies were compared. In the alloy matrix deformed to 75 %, the
distances between the high angle grain boundaries (HAGB) in the normal direction (ND) to
the sheet plane were typically smaller than 1 m, as shown in Fig. 2 c. In the pure metal, the
thickness of similarly oriented layers often exceeded 10 m, even for the deformation of 90
%, as shown in Fig. 1 a, c. These layers were composed of parallel bands or clusters of
subgrains that were strongly elongated in the RD. In some of them, relatively small but
accumulating disorientation angles
1
occurred (Fig. 1c). Because of this accumulation, large
disorientation angles (up to ~ 20 ) between the first and the last subgrain in the band
occurred frequently. Orientation changes of this type are characteristic for transition bands
(Dillamore et al., 1972). The alternation of orientation patterns was also recognized. The high
frequency of the low angle grain boundaries (LAGB) inside the band indicates a well-
developed subgrain structure. With no further analysis of the pure metal deformation
microstructure, we can only conclude that it consists of thick deformation and transition
bands, and the density of HAGBs in the ND is low.
The crystallographic orientation characteristics of the alloy were quite different from those of
the pure metal. The matrix consisted of well-developed HAGBs. The distances between them
along ND were much smaller than those in the pure metal and never exceed a few hundreds of
nanometers. LAGBs in the elongated thin matrix grains were less ordered and created a less-
expanded subgrain structure. In the laminar microstructure of the alloy, small (<<1 m) and
large (13 m) precipitate particles of the second phase were scattered. Around the large
particles, zones of localized strain were identified. These deformation zones consisted of
1
A given misorientation can be described by a rotation axis and an angle of rotation. For a material with
crystal symmetry, there is more than one angle of rotation. The angle with the absolute value smallest of
all possible angles of rotation is called the disorientation angle.
Recrystallization
48
ultrafine (50 200 nm) grains and more or less bent microbands of the matrix (Fig. 3 b, 4 b). The
orientation image of the alloy clearly suggests that microstructural evolution occurs by grain
subdivision at a very small scale compared to the original grain size, which was approximately
100 m in this case. This result is presented in Fig. 2 c and 3 d, showing large orientation
variations over a region as small as a few micrometers. Such rapid changes in orientation may
demonstrate that volumes characterized by a combination of slip systems can be very small.
Fig. 1. As-deformed microstucture of 90% cold-rolled aluminum, the TEM bright field image
(b) and orientation topographies (a and c). On the map (c) the color change indicates a
deviation from the initial orientation (dark blue) to the disoriented orientation (light blue);
black lines (a) - low angle grain boundaries (>1
o
), red lines (c) - the boundaries with a
disorientation angle > 5 ; the points where the diffraction pattern has not been solved are
shown in white.
Application of Orientation Mapping in TEM and SEM for Study of Microstructural
Evolution During Annealing Example: Aluminum Alloy with Bimodal Particle Distribution
49
Fig. 2. a) As-deformed microstucture of 75% cold-rolled 6013 alloy, longitudinal section,
TEM. b) Orientation topography of the matrix area, black lines show high angle grain
boundaries; white regions are not indexed. c) Disorientation angle profile along A-A line (b),
(Sztwiertnia et al., 2007).
The global crystallographic textures of both materials were similar and corresponded to the
well-known rolling texture of FCC metals with high stacking fault energies (such as pure
aluminum). Such a texture is characterized by the concentration of components along two
orientation fibers. The main one, called the fiber, runs diagonally through the orientation
space containing the preferred texture components S {123}<634> and Cu {112}<111>. The
other one, called the fiber, includes Goss {011}<100> and Bs {011}<211> components
(Sztwiertnia et al., 2005).
Recrystallization
50
Fig. 3. a) As-deformed microstucture of 75% cold-rolled 6013 alloy, the deformation zone
around the large particle, longitudinal section, TEM. Orientation topographies in areas of
the deformation zone before (b) and after (c) heating in situ in TEM; particles of the second
phase are shown in black, white regions are not indexed; thick lines indicate high angle
grain boundaries, thin lines indicate low angles grain boundaries. d) Examples of
disorientation angle profiles along E-E and F-F (Sztwiertnia et al., 2007).
Application of Orientation Mapping in TEM and SEM for Study of Microstructural
Evolution During Annealing Example: Aluminum Alloy with Bimodal Particle Distribution
51
2.3 Microstructure changes during annealing
The cold-rolled alloy was tested using the non-isothermal annealing method in a differential
calorimeter. On the basis of these tests, as well as the microscopic analysis of the
microstructures of the appropriately annealed samples, it was possible to state that the two
separate peaks of the stored energy release correspond to the two stages of the
recrystallization process (Fig. 6 a).
The in situ tests in the TEM allowed the determination of the sequence of events occurring at
the beginning of recrystallization in the deformation zones around the large particles and in
the matrix beyond those areas. First, the deformation microstructure was carefully
examined. In the deformation zones around the large second phase particles, small grains
and distorted fragments of microbands were identified. The small grains were
approximately 50-200 nm in size. The strong orientation changes, greater than 15, either
identify HAGBs lying at distances lower than 200 nm or they are an effect of strong grain
bending, as shown in Fig. 3 b, 4 b. To precisely distinguish between these two phenomena,
small-step orientation maps were generated. The broad distribution of orientations in the
deformation zones tended to group in the range of the deformation components after
rotation around the transverse (TD) or the normal direction (ND) to the sheet plane
(Sztwiertnia et al., 2005). This type of rotations suggests that at least a part of the strong
orientation changes may be a result of the accumulation of small disorientations along a
bent grain. In the matrix outside of the DZs, the HAGBs lie roughly parallel to the sheet
plane at distances of 0.5 - 1 m along the ND (Fig. 2 b).
After the sample was annealed in the microscope, the same areas as in the deformation state
were investigated. Figures 3 c and 4 c show examples of DZ orientation maps after in situ
annealing. Nuclei and new grains appeared in the vicinity of the deformation zone. The
orientations of crystallites in the deformed state commonly lay in the area of a particular
new grain or nucleus (Fig. 4 d, e). For each orientation of a new grain, at least one similarly
oriented fragment of the deformed matrix was found. Some of the nuclei were growing
within the zone defined by the migration of HAGBs, a result that was confirmed by a
significant reduction of their density in those zones after annealing to the temperature of the
first peak (Fig. 3 c, d and 4 b, c). The shape of some new grains suggests that they could have
been formed as a result of the local recovery of strongly bent fragments of matrix
microbands, as shown in Fig 4 c.
Partial misorientation distribution functions (PMDF) were calculated between the grains in
the deformed state and the new grains that appeared at the same location (Fig. 5). PMDFs
for both the 75 and 90 % deformed materials show a random distribution of misorientations.
This distribution suggests that there was no special orientation relationship describing
favored growth.
In the deformation zones at very early stages of the recrystallization, broad-spectrum
HAGBs appear to have been active and mobile. At that time, no migration of HAGBs in the
matrix areas outside the zones was observed.
To examine the significance of the in situ experiments, the microstructures of TEM-annealed
thin foils were compared with the microstructures of thin foils prepared from bulk samples
heated in a calorimeter to obtain a specified recrystallization stage. The comparison shows
Recrystallization
52
that the processes occurring in the thin foils and in the bulk samples were, at least
qualitatively, the same. The temperature range of the first recrystallization peak was found
to produce nucleation, which is accompanied by some limited enlargement of new grains in
the sheet plane.
Fig. 4. a) Microstructure of 6013 aluminum alloy cold rolled to 90%, the deformation zone
surrounding a large particle (a), longitudinal section, TEM. Orientation topographies and pole
figures in areas of the deformation zone before (b, f) and after (c, g) heating in situ in TEM;
particles of the second phase are shown in black, white regions are not indexed; thick lines
indicate high angle grain boundaries, thin lines indicate low angle grain boundaries. Areas of
similar orientations (blue, red, yellow) before and after annealing (d, e) (Bieda et al., 2010).
Application of Orientation Mapping in TEM and SEM for Study of Microstructural
Evolution During Annealing Example: Aluminum Alloy with Bimodal Particle Distribution
53
Fig. 5. Partial misorientation distribution functions showing the orientation relationships
between the crystallites from the deformation zones (before annealing) and the new grains
appearing in their positions after annealing (a) 75% cold-rolled 6013 aluminum alloy, (b)
90% cold-rolled 6013 aluminum alloy; Rodrigues representation r
1
, r
2
, r
3
, cross-section
r
3
=const., asymmetric domain (O, O) (Bieda et al., 2010).
Recrystallization
54
By way of example, Fig. 6 c shows the orientation topography in the 75 % cold-rolled sample
that was heated to 330 C in a calorimeter. The appearance of new grains with sizes up to a
few micrometers can be observed around the large particle. These new grains, similar to
those in the in situ experiment, were formed as a result of nucleation and the consequent
growth of nuclei inside the zone. At the examined temperature, the arrangement of LAGBs
can also be observed in the elongated grains of the matrix outside the zone. The annihilation
of these boundaries takes place at higher temperatures, near the end of the first peak.
Fig. 6. a) Power differences, representing the release of stored energy from 75 % cold-rolled
6013 aluminum alloy as a function of annealing temperature. b, c) As-deformed
microstructure and orientation topography of 6013 aluminum alloy 75 % cold-rolled and
subsequently heated in the calorimeter to 330 C: particles of the second phase are shown in
black, white regions are not indexed; thick lines indicate high angle grain boundaries, thin
lines indicate low angle grain boundaries; TEM, (Sztwiertnia et al., 2007).
Application of Orientation Mapping in TEM and SEM for Study of Microstructural
Evolution During Annealing Example: Aluminum Alloy with Bimodal Particle Distribution
55
The phenomenon of the annihilation of LAGBs and the growth of new grains in the sheet
plane (mainly, parallel to the RD) was also observed in the SEM results (Fig. 7). The SEM
measurements of the orientation topographies were made on samples heated to appropriate
temperatures from the range of the first peak (Fig. 5 a). These observations fully confirm the
TEM results and show that after the recrystallization of the deformation zones, local
recovery processes (coalescence of subgrains) take place in the matrix. These local recovery
processes lead to the annihilation of the LAGBs between chains of subgrains lying parallel to
the sheet plane and, consequently, to the production of long grains with a low density of
lattice defects. The growth of the elongated grains in the ND occurs rarely or not at all
within the temperature range of the first peak. The migration of HAGBs in the matrix
becomes the main process within the temperature range of the second peak. Heating of the
sample to the temperature of the end of this peak leads to the complete discontinuous
recrystallization of the material (Fig. 7 c). The recrystallized microstructure is dominated by
elongated grains (up to 30 m in length along the RD). It also includes groups of smaller
grains, which are often almost equiaxial. After the discontinuous recrystallization, the grains
are a few times thicker than they are in the state observed at the end of the first stage.
Fig. 7. Orientation topographies of recrystallized grains in 6013 aluminum alloy 75% cold
rolled and subsequently heated in the calorimeter to: (a) 330 C, (b) 350 C and (c) 480 C,
regions of unsolved diffraction (approximately corresponding to the deformed areas) are
shown in white, EBSD/SEM/FEG. d) and e) Schematic representation of the deformed
microstructure before and after annealing to the temperatures from the end of the first
recrystallization peak, Fig. 2; thick lines indicate high angle grain boundaries, thin lines
indicate low angle grain boundaries.
3. Summary
Orientation topography provides basic local and global information about microstructures
by allowing the identification and description of occurring regularities. The OM technique
in TEM is a useful tool for the quantitative and qualitative characterization of fine crystalline
and deformed microstructures in polycrystalline materials. It is possible to obtain
information about grain distribution, misorientation between grains, material phases, the
local orientation distribution function, and the misorientation distribution function.
Replacement of the SEM measurements by TEM measurements improves the spatial
resolution to a few nanometers. Both SEM and TEM can be used for complementary analysis
of crystalline materials at the micro and nano scale, respectively. Together with in situ
studies, orientation mapping in TEM can provide additional information about the behavior
Recrystallization
56
of a material during annealing, particularly in zones of greater deformation. The example of
orientation characteristics presented in this chapter illustrates only some aspects of the
applicability of these techniques. Orientation mapping in a transmission electron
microscope was successfully applied to the study of microstructural changes during the
initial stage of recrystallization in an aluminum alloy with a bimodal second-phase particle
distribution. The images of the microstructure in the representative areas of a sample of
deformed aluminum alloy 6013, described by measurements of orientation topography,
shows greatly advanced grain fragmentation.
In situ investigations in TEM, calorimetric measurements, and orientation mapping in TEM
and SEM demonstrate that the recrystallization of the tested material can be considered to
occur as a number of partly overlapping processes that proceed in two stages. These stages
correspond to the two separate stored energy release peaks. In the initial stage, the
deformation zones around large second phase particles act as sites for particle-stimulated
nucleation. This nucleation is accompanied by the growth of nuclei. However, the migration
of high angle grain boundaries only occurs in deformation zones. At the same range of
temperatures, some enlargement of new grains in the matrix (outside of the deformation
zones) was also observed. The formation of grains elongated primarily in the direction
parallel to the rolling direction may be correlated to the processes of local recovery, which is
triggered in the deformation zones. Grain elongation then continues to develop along the
bands of the deformed matrix in the directions of low orientation gradients. The elongated
grains appear because of the annihilation of low angle grain boundaries between chains of
subgrains lying in layers parallel to the sheet plane. As a consequence, new grains often
have a plate-like character, with their shorter axis parallel to the sheet plane normal
direction. Their lengths along the rolling direction may exceed 50 m, while their thickness
corresponds approximately to the distance between high angle grain boundaries in the
normal direction outside the deformation zones; this dimension was not observed to exceed
a few micrometers. In the second stage, high angle grain boundaries were observed to
migrate in the direction of the high orientation gradient. This migration, mostly in the
normal direction, was limited to "free areas" of the deformed matrix between bands of new
grain formation in the initial stage of recrystallization.
4. References
Adams B.L., Dingley D.J (1994), Orientation Imaging Microscopy: New Possibilities for
Microstructural Investigations using Automated BKD Analysis, Mater. Sci. Forum, 157-
62 31.
Ardakani M.G., Humphreys F.J. (1994), The annealing behavior of deformed particle-containing
aluminum single crystals. Acta Metal. Mater., 42, 763.
Bailey J.E. (1960), Electron microscope observations on the annealing process occurring in cold-
worked silver, Phil. Mag., 5, 833.
Bieda M.,.Sztwiertnia K, Korneva A., Czeppe T., Orlicki R. (2010), Orientation mapping study
on the inhomogeneous microstructure evolution during annealing of 6013 aluminum alloy,
Solid State Phenom, 16, 13.
Application of Orientation Mapping in TEM and SEM for Study of Microstructural
Evolution During Annealing Example: Aluminum Alloy with Bimodal Particle Distribution
57
Bieda-Niemiec M. (2007), Opracowanie systemu do automatycznego pomiaru map orientacji w
transmisyjnym mikroskopie elektronowym do analizy mikrostruktury drobnoziarnistych
materiaw metalicznych, Thesis, Krakw IMIM PAN, in polish.
Dillamore I.L., Morris P.L., Smith C.J.F., Hutchinson W.B. (1972), Transition Bands and
Recrystallization in Metals, Proc. Roy. Soc., 329A, 405.
Dingley D.J. (1984), On-line determination of crystal orientation and texture determination in
SEM, Proc. Roy. Microsc. Soc., 19, 74.
Dingley D.J. (2006), Orientation Imaging Microscopy for the Transmission Electron Microscope,
Mikrochim. Acta, 155, 19.
Haessner, F., Pospiech, J. and Sztwiertnia, K. (1983), Spatial arrangement of orientations in
rolled copper, Mat. Sci. Eng.1, 1.
HKL (2007) http://www.oxford-instruments.com
Hu H. (1963), Electron Microscopy and Strength of Crystals, Interscience, London, 564.
Humphreys F.J. and Hatherly M. (2002), Recrystallization and Related Annealing Phenomena,
Pergamon Press, Oxford.
Hutchinson W.B., Ray R.K. (1973), On the feasibility of in situ observations of recrystallization in
the high voltage microscope, Phil. Mag., 28, 953.
Morawiec A. (1999), Automatic orientation determination from Kikuchi patterns, J. Appl. Cryst.
32, 788.
Morawiec A. (2004), Orientations and Rotations. Computations in Crystallographic Textures.
Berlin, Heidelberg, New York: Springer-Verlag.
Morawiec A., Fundenberger J.J., Bouzy E., Lecomte J.S. (2002), EP-a program for determination
of crystallite orientations from TEM Kikuchi and CBED diffraction patterns J. Appl.
Cryst., 35, 287.
Pospiech J., Lcke K. and Sztwiertnia K. (1993), Orientation Distribution and Orientation
Correlation Functions for Description of Microstructures, Acta Metall. Mater. , 41, 305.
Rauch E.F., Dupuy L. (2005), Rapid spot diffraction patterns identification through template
matching, Arch. Metall. Mater., 50, 87.
Roberts W., Lehtinen B. (1972), On the feasibility of in situ observations of recrystallization in the
high voltage electron microscope, Phil. Mag., 26, 1153.
Schwartz A.J., Kumar M.,. Adams B.L, Field D.P. (2009), Electron Backscatter Diffraction in
Materials Science, ISBN 978-0-387 88135-2, Springer.
Sztwiertnia K. (2008), On recrystallization texture formation in polycrystalline fcc alloys with low
stacking fault energies, Int. J. Mater. Res., 99, 178.
Sztwiertnia K., Bieda M., Korneva A., Sawina G. (2007), Inhomogeneous microstructural
evolution during the annealing of 6013 aluminium alloy, Inynieria Materiaowa, Vol. 3
XXVIII, 476.
Sztwiertnia K., Bieda M., Sawina G. (2006), Determination of crystallite orientations using TEM.
Examples of measurements, Arch. Metall. Mater. , 51, 55.
Sztwiertnia K., Haessner F. (1994), In situ observations of the initial stage of recrystallization of
highly rolled phosphous copper, Mater. Sci. Forum, 157-162, 1069.
Sztwiertnia K., Morgiel J., Bouzy E. (2005), Deformation Zones and their Behaviour during
Annealing in 6013 Aluminim Alloy, Arch. Metall. Mater. 50, 119.
TSL (2007) http://www.edax.com
Recrystallization
58
Wright S.I., Adams B.L. (1992), Automatic analysis of electron backscatter diffraction patterns,
Met. Trans. A 23, 759.
Zaefferer S, Baudin T, Penelle R (2001), A study on the formation mechanisms of the cube
recrystallization texture in cold rolled Fe-36% Ni, Acta Mater. , 49, 1105.
4
Crystal Growth:
Substructure and Recrystallization
Vadim Glebovsky
Institute of Solid State Physics, the Russian Academy of Sciences
Russia
1. Introduction
Single crystalline refractory transition metals (molybdenum, tungsten, niobium, and
tantalum) exhibit a unique combination of properties, namely, high strength, plasticity,
Young modulus, wear resistance, number, and low coefficient of linear expansion as well
as high radiation resistance which is what makes single crystals of these metals of high
purity the most suitable materials to be widely employed in science and engineering.
Single crystals of high purity tungsten are well suited to production of the deflectors of
the charged particles beams in the linear accelerators, the colliders and, also, to be
successfully used as the target-converters for the sources of the positron beams. The
attributes of single crystals of molybdenum alloys, compared to their polycrystalline
counterparts, include more stable microstructures, lower creep rates, better compatibility
with nuclear fuels and lower diffusion penetrability (Liu & Zee, 1996). On the other side,
studies of the X-ray wave field in crystals or so called effects of the dynamic scattering
theory are of high interest although the first observations of X-ray anomalous
transmission are made more than fifty years ago. The necessary high degree of structural
perfection is achieved for a limited number of crystals, such as silicon, germanium, and
related families; almost no observations of the dynamic effects have been made in metals.
Studies of X-ray anomalous transmission in the transition metals are of considerable
interest, particularly in tungsten which has a simple structure and a high absorption
coefficient.
A method of electron-beam floating zone melting (EBFZM) is widely used to grow single
crystals of high-purity refractory transition metals for years (Pfann, 1966; Shah, 1980).
The growth of the perfect single crystals of the refractory metals presents difficulties
because of the low defect formation energy and the stringent constraints on the level of
the temperature gradients. The single crystals of molybdenum and tungsten, grown from
the melt by this method, tend to have the specific substructure, characterized by the high
dislocation density, reaching 10
5
-10
7
cm
-2
. The main part of these dislocations is collected
in the walls, forming the dislocation substructure of the three orders of magnitude. On
the one hand, the substructure is due to polygonization of dislocations arising during the
growth by one of the known mechanisms (Bolling & Finestein, 1972, Kittel, 1996, Nes &
Most, 1966). On the other hand, there is inevitable inheritance of the seed crystal
substructure in the growing single crystal, which consists in the fact that the favorably
Recrystallization
60
oriented low-angle boundaries grow up into a crystal. Considerable efforts have
therefore been made to improve the structural quality of the tungsten single crystals
(Cortenraad et al., 2001a). Modern methods of preparing the tungsten single crystals can
produce the specimens having the dislocation density of about 10
5
cm
-2
. A chemical
composition, a growth rate, a number of passes by the liquid zone, geometry of the
crystals and some other parameters of the growth in varying degrees affect the
substructure, but in any case, the substructure of the crystals of molybdenum and
tungsten grown from the melt is far imperfect (Glebovsky et al., 1988; Glebovsky &
Semenov, 1993-1994, 1995, 1999)
However, the single crystals, free of the specific substructure, can be grown by the
secondary recrystallization process, which consists of the plastic deformation procedure and
the high-temperature annealing procedure. The plastic deformation procedure of
monocrystalline specimens can be produced by rolling in the vacuum rolling machines. The
high temperature annealing procedure can be performed with the help of the anneal devices
inside the rolling machines or in the EBFZM set-ups. The studies of structural perfection of
the single crystals grown from the melt and by recrystallization are made by using the
methods of X-ray rocking curves and angular scanning topography. To monitor the
subgrain substructure of the tungsten single crystals, the X-ray anomalous transmission
method has been employed as well. The optimal recrystallization process involves the
deformation of single crystals with the [111] growth axis by rolling along the (112) plane.
The 6-12% deformation is found to be optimal to get the polycrystals with the large grains of
high perfection. The vacuum conditions are most suitable for vacuum rolling to avoid
oxidation of the crystal surfaces during deformation at high temperatures. As a result, the
single crystals of molybdenum and tungsten have the substructure which is characterized
by both the record-low dislocation density and the small mosaic (Glebovsky & Semenov,
1999). For comparison, the tungsten single crystals, grown from the melt, contain the
subgrains of the first order, elongated along the growth axis with the misorientation angles
of 8-10 of an arc. The crystallographically perfect tungsten single crystals, obtained by
recrystallization, do not contain the subgrains of the first and second orders at all, and the
maximum misorientation angles of the subgrains of the third order are less than 1 of an arc.
The structural changes in the perfect single crystals as a result of the thermal stresses, when
they have been used as the seed crystals for growing the single crystals from the melt, have
been studied.
It is well known that ideal growth techniques and technologies do not exist. All methods
and technologies have their own advantages and disadvantages, so the main tasks of
researchers consist in developing the advantages and in reducing negative effects of the
disadvantages. The EBFZM method has its unique advantages which open wide prospects
in the production of the high-purity refractory metals. It would be a mistake if the prospects
will not be realized because of the complexities associated with the structural features of the
single crystals grown from the melt.
The idea of the chapter is to show the most reliable ways of improving the structural quality
of single crystals of the high-purity refractory metals. The recrystallization example for the
tungsten single crystals shows how perspective and reliable are these ways in obtaining the
structurally perfect single crystals.
Crystal Growth: Substructure and Recrystallization
61
2. Brief comments on the electron beam float-zone melting and growing
single crystals
Zone-melting techniques (particularly, the EBFZM method) are useful for metals which are
very reactive in the liquid state at high temperatures, so they cannot be processed at any
contact with other materials (Pfann, 1966). Quoting Pfann, discovered the zone melting
technique: I regard the conception and development of zone melting as an exiting scientific
advance. And I cannot help being saddened to hear it occasionally referred to as simply a
technical innovation that was mysteriously evoked by the need for transistor grade
germanium and silicon. I regard zone melting as elegant both in its simplicity and its
surprising complexity. By far the EBFZM method, which is the crucibleless zone melting
technique, is characterized by simplicity and complexity, but it is still the best one for
melting refractory metals and their alloys. There are some well-known advantages of the
method: small volume of the melt; the high efficient EB guns; the well-defined thermal
gradients; no contamination from the crucible materials because instead of them the
surface tension of liquid metals works; the non-contaminating way of heating - the electron
beams; effective purification which can be achieved due to evaporation of impurities in a
vacuum. In parallel, there are some disadvantages of the EBFZM method: it can only be
used in a vacuum; limitation of sizes of crystals due to surface overheating and thus
decreasing the surface tension of the liquid zone; the high axial temperature gradients in a
solid; the high thermal stresses and the high density of the unremovable dislocations;
limitation of geometry and mass of the crystals produced by the EBFZM method. These or
other properties of the method are marked as the merits or shortcomings, it would be wrong
to perceive clearly. Thus, one of the major advantages of the method is absence of refractory
crucibles and holding the liquid zone by the surface tension. However, high sensitivity of
the surface tension to the surface-active impurities and the temperature gradients converts
the recognized advantage into a serious drawback, which prevents growing the single
crystals of large diameters, because mass of the liquid zone is too large so that surface
tension forces are able to hold it. A similar comment can be done concerning the
temperature distribution and the temperature gradients. Certainly, the well-defined thermal
gradients are the advantage of the method, but their high values lead to formation of the
specific substructure in single crystals, which creates great problems for both the physics
research and industrial application. The electronic heating is the really controlled non-
contaminating way of heating, but it may only be used in a vacuum, which is also a kind of
contradiction when discussing the advantages and the disadvantages of the method.
3. Features of the single crystals growth
The single crystals of the high-purity refractory metals are widely used in modern material
science and technology (Alonzo et al., 1995; Calverly et al., 1957; Glebovsky et al., 1998; Hay
et al., 1968; Liu & Zee, 1996; Moest et al., 1998). This necessitates both studying purification
processes and developing advanced techniques of growing single crystals of high-purity
refractory metals with modern electron beam (EB) guns (M. Cole et al., 1968; Glebovsky et
al., 1986). Crucibleless techniques with electronic heating are extremely important for
melting, studying and preparing refractory metals because of their high chemical reactivity
in the liquid state. Early zone refining theories such as progressive freezing, zone refining,
zone crystal growing, nonideal separation and optimization are studied and discussed
Recrystallization
62
elsewhere (Pfann, 1966; Shah & Wills, 1975). Equipment and technologies such as types of
the electron guns used, the drive mechanisms, heating and cooling, floating-zone melting,
and stirring are also discussed in detail elsewhere (Shah, 1980). In the EBFZM method the
liquid zone is held in place between two vertical collinear solid rods by its surface tension
(Fig. 1). Single crystals of high-purity refractory metals can be grown exclusively by EBFZM
because of their extremely high melting temperatures and chemical reactivity (Calverly et
al., 1957; Hay et al., 1968; Alonzo et al., 1995; Moest et al., 1998; Glebovsky et al., 1998). This
necessitates both studying purification processes and developing advanced methods of
growing single crystals metals using modern electron beam guns (M. Cole et al., 1968;
Glebovsky et al., 1986). The main purpose in this field is to study the real structure of single
crystals as a function of the technological parameters of the EBFZM method (Langer, 1980;
Riedle et al., 1994, 1996).
Fig. 1. Thermal zone of the EBFZM.
For effective melting and growing, the original EB guns have been elaborated on because
the EB guns is the most important element of the EBFZM set-ups (Glebovsky et al., 1986;
Shah, 1980). In Fig. 2 the EB gun is shown, which consists of a cathode, an anode, and the
focusing electrodes. The main features of the EB gun are: (a) the rod (crystal) serves as the
anode, (2) the focusing electrodes made of a water-cooled copper, which makes the EB gun
geometrically solid even at very high temperatures in the liquid zone, (3) the focusing
electrodes form a stable circular electron beam field and focus it on the liquid zone, (4) the
EB gun produces the well-defined thermal gradients on the crystal under the crystallization
front. The cathode is made of a circular tungsten filament of 55 mm in dia. An arrangement
of the focusing electrodes makes it possible to vary the electron-beam field from a diffuse
pattern to sharp one. The advantage of the EB gun is its effectiveness at the refining and
growing procedures during service of about 200 hours, compared to the known EB guns
which can be used for no longer than 20-30 min. Thus, the original EB gun can be used
Crystal Growth: Substructure and Recrystallization
63
continuously, both for refining of refractory metals and growing the single crystals at the
growth rates of up to 50 mm/min, diameters up to 35 mm and lengths up to 1100 mm
(Glebovsky et al., 1986). The growth of single crystals is usually accompanied by purifying
liquid metals to high purity. It is demonstrated by preparation of the high-purity refractory
metals with the residual impurities at the level of detection of the modern analytical
techniques (Alonzo et al., 1995; Bdikin et al., 1999; Bozhko et al., 2008; Chaika et al., 2009;
Brunner & Glebovsky, 2000a, 2000b; Cortenraad et al. 2001a, 2001b, 2001c, 2001d; Ermolov et
al., 1999, 2002; Glebovsky et al., 1998; Markin et al., 2006, 2010; Moest et al., 1998; Shipilevsky
& Glebovsky, 1989). The most problematic metals in growing the single crystals of the
refractory metals (molybdenum, tungsten, niobium, and tantalum) are two - molybdenum
and tungsten. Therefore, the focus of this chapter is devoted to just these two metals,
although almost all the results can be easily applied to other two metals - niobium and
tantalum.
Fig. 2. Circular EB gun with focusing electrodes made of water-cooled copper.
4. Substructure of the molybdenum and tungsten single crystals
Single crystals of refractory metals with the relatively simple bcc-lattice grown by EBFZM
have the specific dislocation substructure with the size of subgrains, which can be divided
into three orders of magnitude. Table 1 shows approximate parameters of the substructures.
The chemical composition, especially the content of the interstitial impurities, the growth
rate, the number of the liquid zone passes, geometry of the crystal and other parameters
significantly affect structural perfection of crystals.
There are several mechanisms of appearance of both the dislocations and substructure of
single crystals at growing from the melt (Hurle, 1977; Reid, 1966): under influence of
thermal stresses during growing and cooling of single crystals, due to the impurity
concentration gradients in the solid phase, due to supersaturating of the lattice with
vacancies, inheritance of the substructure of the seed crystal into the growing crystal. In
reality, apparently, several mechanisms can operate simultaneously, or some of them will
dominate. Obviously, in single crystals of the sufficiently pure refractory metals, effect of
impurities on the substructure is unimportant (Akita et al., 1973). However, despite of a
large number of studies in this area still remain unclearness related with influence of some
factors in formation of the substructure.
Recrystallization
64
Order of substructure Average size of subgrains
Misorientation angles
between subgrains
First order 1 mm < d < 8 mm 30 < < 4
0
Second order 50 m < d < 1 mm 30 < < 30
Third order 0 < d < 50 m 0 < < 30
Table 1. Estimated parameters of the crystalline substructure.
In growing the single crystals of molybdenum and tungsten by EBFZM one of the main
monitored parameters is the growth rate (or the rate of liquid zone traveling, or the rate of
the EB gun displacement). The growth rate is essential both at the crystallization stage, and
at the post-crystallization annealing stage, which begins at an interface between the liquid
zone and single crystal. At high temperatures, dislocations, regardless of nature of their
origin, have very high mobility, so that there polygonization of the dislocation substructure
has taken place. Apparently, along with increased dislocation mobility the stresses at the
interface between the solid and liquid phases also contribute to formation of the
polygonized structure. The typical substructure of the tungsten single crystal with the
growth axis [001], revealed at the (010) plane parallel to the growth axis, is shown in Fig. 3.
It is clearly seen that the boundaries of the subgrains of the first order extend for the
considerable distances along the growth axis. They represent the walls or the dislocation
network formed by potential for the bcc-lattice dislocations with the Burgers vectors b
1
= a/2
[111] and b
2
= a [100].
Fig. 3. Substructure of the tungsten single crystal with the vertical growth axis [001] grown
at the rate of 2 mm/min. Electrolytic etching in 25% solution of NH
4
OH.
The impurities and doping have significant effect on the substructure of single crystals
grown from the melt. When microalloying occurs at the definite growth rate, the flat
crystallization front is quite stable, and concentration supercooling does not develop.
Changes in the dislocation substructure are associated with increase of the dislocation
density inside subgrains, decrease of the average size of subgrains and increase of the
misorientation angles between subgrains. In the case of doping in significant concentrations,
the growth of single crystals becomes impossible at any growth rate. Figure 4 shows the
longitudinal and transverse cross-sections of the polycrystalline Mo-2%W alloy, grown from
the melt by EBFZM. This polycrystalline ingot contains the large grains elongated along the
growth axis. Figure 5 shows the microstructure of the polycrystalline molybdenum ingot of
low purity, also grown from the melt by EBFZM. The resulting structure is distinctive: the
single-crystalline core surrounded with the polycrystalline periphery.
Crystal Growth: Substructure and Recrystallization
65
Fig. 4. Microstructure of the polycrystalline Mo-2%W alloy, grown from the melt by EBFZM
(longitudinal and transverse cross-sections).
Fig. 5. Microstructure of the polycrystalline molybdenum ingot of low purity, grown from
the melt.
To study effect of the growth rate on the substructure, the single crystals of molybdenum
and tungsten, having the growth axes along the [001] axis, are grown. The growth rate
varied from 0.2 to 50 mm/min, the rotation rate of the growing crystal - from 0 to 100
revolutions per minute. The range of the growth rates under study actually covers all
growth rates, implemented in the EBFZM method: the lower limit depends on intense metal
evaporation, and the upper limit by possibility of full melting and stability of the liquid
zone. The single crystals are grown in three passes of the liquid zone: the first pass at the
growth rate of 6 mm/min, the second pass at the growth rate of 2 mm/min, the third pass -
on the seed crystal at the rate from the above range of the rates.
4.1 The substructure of the molybdenum single crystals
The substructures of the molybdenum single crystals vary seriously depending on the
growth rate (Glebovsky et al., 1988; Glebovsky & Semenov, 1994, 1995; Glover et al., 1970,
Liu & Zee, 1996). At high growth rates the substructure is characterized by the high
dislocation density and the more stressed state. The latter is confirmed by zone annealing
the molybdenum single crystal of 8 mm in dia at the temperature close to the melting
temperature and at the traveling rate of the EB-gun of 0.5 mm/min. At the low growth rates
the dislocations have enough time to be polygonized, because single crystals stay at elevated
temperatures for a longer time. Studies of the molybdenum single crystal grown at the
growth rate of 6 mm/min show that after zone annealing the crystal has been polygonized
to greater extent.
Recrystallization
66
The single crystals grown at the growth rate of 0.5 mm/min have the specific developed
substructure with an average size of subgrains of the second order of about 100 m and the
dislocation density, calculated from the etch pits, of 3x10
5
cm
-2
. The substructure of the
single crystal grown at 6 mm/min is characterized by the individual etch pits and the lack of
the polygonized boundaries. The dislocation density is higher - of 1x10
6
cm
-2
. The specific
substructures of the molybdenum single crystals grown at different growth rates are shown
at Fig. 6, 7, and 8. Most clearly the dependence of the misorientation angles of the subgrains
on the growth rate is detected by the divergent X-ray beam patterns (Fig. 9). The
misorientation angles of the subgrains of the second order in accordance with the value of
discontinuities on the line (400) is 50 of an arc for the growth rate of 0.5 mm/min, and 20 of
an arc for the growth rate of 6 mm/min.
Fig. 6. Specific substructures of longitudinal (a) and transverse (b) cross-sections of the
molybdenum single crystal, grown at 2 mm/min.
Fig. 7. Specific substructures of longitudinal (a) and transverse (b) cross-sections of the
molybdenum single crystal, grown at 10 mm/min
Fig. 8. Specific substructures of longitudinal (a) and transverse (b) cross-sections of the
molybdenum single crystal, grown at 40 mm/min.
Crystal Growth: Substructure and Recrystallization
67
Fig. 9. Divergent X-ray beam patterns of the molybdenum single crystals, grown at rates: a-
0.5 mm/min, b- 6 mm/min.
Another feature of the substructure of the single crystals of molybdenum is radial
heterogeneity, detectable at all growth rates. In the central part of the single crystals the
boundaries are almost absent; however, at the periphery is observed intense polygonization.
Radial inhomogeneity of the single crystals of 8 mm in dia is clearly visible on the divergent
X-ray-beams patterns, where the value of shifts at edges of the line (310) is bigger than in the
central part: at the edges 50 of an arc, in the center 20 of an arc. Dependence of nature of
the substructure of the single crystals on radial inhomogeneity as well as on the growth rate
can be explained by dislocation motion in the non-uniform temperature field of the single
crystals.
4.2 The substructure of the tungsten single crystals
The typical substructures of the tungsten single crystals with the growth axis [100] and 10-
12 mm in dia, grown at growth rates from 0.5 mm/min up to 40 mm/min, are shown in
Fig. 10, 11, and 12. At the growth rates of 0.5-4 mm/min on the transverse cross-sections
in the (001) plane, the distribution of the subgrains of the second order by size is close to
normal. On the longitudinal cross-sections in the plane (100), pronounced elongated
subgrains along the growth axis are found. At 40 mm/min, the polygonization process is
not completed: the dislocation density inside the subgrains increases. At two different
sites of the tungsten single crystal grown at the growth rates of 2 mm/min and 40
mm/min, the dislocation density increases by a half of an order of magnitude - from 8x10
4
cm
-2
to 2x10
5
cm
-2
. There is also decrease of the average size of the subgrains, and the
misorientation angles are increased to 3-4
0
of an arc. Apparently, such influence of the
growth rate on the substructure of the tungsten single crystals is common for all
refractory transition metals.
Fig. 10. Specific substructures of longitudinal (a) and transverse (b) cross-sections of the
tungsten single crystal grown at 0.5 mm/min.
Recrystallization
68
Fig. 11. Specific substructures of longitudinal (a) and transverse (b) cross-sections of the
tungsten single crystal, grown at 2 mm/min.
Fig. 12. Specific substructures of longitudinal (a) and transverse (b) cross-sections of the
tungsten single crystal, grown at 40 mm/min.
Note significant irregularity of the substructure of the tungsten single crystals in the radial
direction: the central part and the periphery are notably different, similar to differences
observed in the molybdenum single crystals. The metallographic studies of the substructure
of the single crystals of molybdenum and tungsten show that between them there is
fundamental similarity. The rate of rotation, and the focusing and power fluctuations of the
electron beam, leading to emergence of the constrictions and other defects on the surface of
the single crystals, generally have no noticeable effect on the substructure of the single
crystals. The fact that the temperature field is non-uniform in the radial direction has been
confirmed by existence of the curvilinear crystallization front in vicinity of the seed during
zone melting (Fig. 13). In turn, heterogeneity of the temperature field leads to non-uniform
mechanical stresses which have different effects on the dislocation motion rate.
Fig. 13. Macrostructure of the longitudinal thin section of the tungsten single crystal of 11
mm in dia, showing the curvilinear crystallization front in vicinity of the seed.
Crystal Growth: Substructure and Recrystallization
69
Although unevenness of the substructure of single crystals creates some problems for
physical studies and practical applications (Markin et al., 2006, 2010; Mundy et al., 1978),
there are recent results on successful use of the as-grown single crystals for the manufacture
of the STM tips (Bozhko et al., 2008; Chaika et al., 2009). It is demonstrated the main
advantage of the single crystalline W[001] STM tips: sharpness, stability, and the predictable
atomic structure. With these tips a set of the complimentary atomically resolved images of
the complicated Si(557)5x5 stepped surface reconstruction is reproducibly received and
revealed its atomic structure. The example of instability of the W[001] tip illustrates how the
known tip axis orientation and the apex atom jump lengths may allow one to predict the
atomic structure of the real single crystalline tip that can be of high importance for correct
interpretation of the ultimately high resolution STM data. Nevertheless, presence of the
specific substructures in single crystals, which prevents expansion of the practical
application of single crystals, while insoluble problem for the researchers, that stimulates
search for ways of obtaining the perfect single crystals of these metals.
5. Effect of thermal stresses on the substructure of the single crystals
The EBFZM method is characterized by presence of the high axial temperature gradients,
especially near the solidification front in both the solid and liquid phases. Since the
temperature gradient is a nonlinear function of the distance from the solidification front, i.e.,
d
2
T/dz
2
0, this leads to thermal stresses in growing single crystals that can cause
multiplication of the dislocations. For a cylindrical crystal with dissipated radiating
crystallization heat, the axial temperature gradients can be estimated from the known
formulas relating the temperature, the black-body coefficients, the Stefan-Boltzmann
constant, the crystal diameter, and thermal conductivity (Table 2).
Metal
The distance from the liquid zone Z, cm
0 0,2 0,4 0,6 0,8 1,0 1,2
Molybdenum -618 -557 -505 -461 -422 -389 -359
Tungsten -1453 -1207 -1021 -877 -763 -671 -596
Table 2. Axial temperature gradients (K/cm) in the solid state for single crystals of
molybdenum and tungsten
The temperature along the axis of single crystals has been measured by optical
micropyrometry. The holes of 1 mm in dia and 7-8 mm in depth, located along the axis of
the single crystals, serve as a black-body model. Due to the high axial temperature
gradients, the black-body model is appeared to be essentially non-isothermal. This
increases the temperature measurement error up to 100
0
, but it is still possible to obtain
the reproducible temperature profiles for all metals studied. Figure 14 shows the
temperature distribution along the axis of the cylindrical tungsten single crystal of 15 mm
in dia. It is seen that near the crystallization front the temperature along the crystal falls
particularly sharply to 2000K at 40 mm from the front. The control of the temperature
profile along the axis of the liquid zone is performed by measuring an electron current.
Owing to the design of the EB gun, the temperature profile of the liquid zone can be
effectively changed from diffuse one to sharp, which gives additional opportunity to
manage the crystal growth.
Recrystallization
70
Fig. 14. Temperature distribution along the axis of the cylindrical tungsten single crystal of
15 mm in dia.
For the case of the ax-symmetric temperature distribution along the cylindrical single
crystal, resulting thermal stresses are mainly determined by the axial temperature gradients.
At pre-melting temperatures, an elastic limit of metals is practically zero and thermal
stresses are completely removed by the dislocations, i.e., there is plastic deformation. Since
at the growth of crystals from the melt the crystallization heat should released, then
inevitably there is the temperature gradient in the solid phase, which leads to the certain
density of unremovable dislocations (M. Cole, et al., 1961; Buckley-Golder & Hurphreys,
1979; Esterling, 1980; Nes & Most, 1966; Otani, 1984;). The estimate of this dislocation
density can be performed by:
o > gradT / b, (1)
Here, a linear coefficient of thermal expansion, K
-1
; b - the Burgers vector, cm; gradT - the
temperature gradient at the crystallization front, K/mm. For tungsten single crystals, such
evaluation reveals that in order to have the dislocation density of = 10
4
cm
-2
, the
temperature gradient must be less than 50 K/cm. Since the real temperature gradients are
usually higher for an order of magnitude, in the melt-grown tungsten single crystals the
dislocation density is typically 10
5
-10
6
cm
-2
or even more. The dislocation density in the
boundaries is usually for an order of magnitude higher than in the bulk of the subgrains.
The estimate by means of the formula (1) with the temperature gradient of 100 K/mm gives
the dislocation density of 5x10
5
cm
-2
(Table 3).
Dislocation density , cm
-2
Temperature gradients, /
Molybdenum Tungsten
10
6
2900 3600
10
5
290 360
10
4
29 36
Table 3. Numerical estimates of temperature gradients at the crystallization front for single
crystals of molybdenum and tungsten of 16 and 11 mm in dia, respectively.
Crystal Growth: Substructure and Recrystallization
71
Note that dislocations in single crystals are formed under action of thermal stresses in the
growth process, and in the cooling process as well. Depending on the cooling rate, the
number of the imposed dislocations in the growing crystal can be even higher than the
number of the dislocations appeared during the growth (Nes & Most, 1966). The result is in
decreasing subgrains sizes, as observed in the thermal shock at welding up both the seed
crystal and initial rod together or cutting the crystal by the electron beam. In addition to the
axial temperature gradients, quantitative estimates for the stationary stage of the crystal
growth have shown, that to form the substructure, the cooling rate is important as well,
which is realized in the growing and cooling processes when the crystal growth process is
over. For example, in aluminum single crystals obtained by zone melting, the residual
dislocation density is of ~ 10
2
cm
-2
when cooled to room temperature at ~ 10
-3
K/s (about a
week), while at the increasing cooling rates on an order of magnitude, the residual
dislocation density increases to ~ 10
4
cm
-2
.
In the case of growing the single crystals of molybdenum and tungsten by EBFZM, one can
get the estimate from below for the maximum cooling rate of the crystal growth process. The
maximum of this magnitude occurs in the solid phase just below the crystallization front.
Multiplying the axial temperature gradient on the growth rate, one obtains the value of the
maximum cooling rate of the crystal during the growth:
.
dt dz dT
dz dt dt
= (2)
Taking the data of the temperature gradients (Table 3) and the growth rate of 2 mm/min,
the most frequently used in practice, one can obtain the cooling rate for molybdenum and
tungsten, respectively, 2K/s and 5 K/s, i.e., the cooling rates are very high, taking into
account the corresponding values for aluminum (10
-3
K/s). It should be noted that
significant reducing the crystallization rate in the EBFZM method is impossible. The fact
that molybdenum and tungsten have very high vapor pressure at T > T
m
and at the growth
rate of 0.5 mm/min, the metal losses by evaporation can reach 30% of initial mass.
However, even if the crystallization rate tends to zero, desired reduction of the dislocation
density still can not be achieved. The reason is that the crystallization rate never coincides
with the growth rate (or the rate of the EB gun displacement). In fact, the crystallization rate
affects significantly the hydrodynamic processes developing in the liquid zone (Kobayashi,
1970; Kobayashi & Wilcox, 1982; Murphy, 1987; Surek & Chalmers, 1975). These processes
give rise to oscillations in the growth rate; moreover, the instantaneous crystallization rate in
these moments of time can be significantly greater than the cooling rate, as shown by
estimates for molybdenum and tungsten. Presence of such oscillations of both the
temperature and growth rate is shown elsewhere (Mullins & Sekerka, 1964; Wilcox & Fuller,
1965). The frequency of these oscillations is close to the inverse of the thermal time constant
of the melt-crystal system:
a
f
S
~ (3)
Here, f a frequency of oscillation, Hz; a a thermal diffusivity, cm
2
/s; S a cross-sectional
area of the crystal, cm
2
. Oscillations of the crystallization front in presence of impurities in a
Recrystallization
72
crystal lead to substantial change in the distribution coefficient and, consequently, the so-
called transverse striations in the crystals observed by autoradiography.
Considerable interest represents an estimate of the cooling rate from T
m
to room
temperature. The simplest case can be considered, when the one-dimensional quasilinear
heat conductivity equation with the constant coefficients is numerically solved. The process
of heat propagation in a homogeneous rod can generally be described by the equation:
' ( , )
p
T
K
T X
C f x t
x
t
c | |
c
|
c c
\ .
= +
c c
(4)
where T (x, t) a temperature at the point X of the rod at the moment of time t; C
P
heat
capacity per unit mass at constant pressure;
m
metal density; K thermal conductivity; f
density of the heat sources (sinks); X a coordinate along the rod length L. If one assumes
that K, C
P
, and
c c
=
c c
(6)
In the task the most interesting is the value of T/t with the limitations 0XL; and t=0.
This value is calculated numerically, and it provides the cooling rate at the upper limit. For
these calculations, the following numerical values of the physical parameters are used. For
molybdenum: K = 0.909 Wcm
-1
K
-1
, C
P
= 0.235 Jg
-1
K
-1
,
Mo
= 10.2 gcm
-3
, d = 1.6 cm. For
tungsten: K = 0.945 Wcm
-1
K
-1
, C
P
= 0.172 Jg
-1
K
-1
,
W
= 19.2 gcm
-3
, d = 1.1 cm. The cooling rate
for molybdenum is 2x10
4
K/s, and for tungsten 5x10
4
K/s. Although these values exceed
the cooling rates at the stationary stage of the crystal growth, they can not significantly
impact on deterioration of the substructure, because in 5-10 seconds for the most part of the
crystal they become comparable with the cooling rates at the stationary phase. The
dislocation density can increase only a few in the surface layer near the end of the crystal.
Therefore, when the crystals of refractory metals are grown from the melt, it is absolutely
impractical to cool slowly, from scientific or technological points of view.
Crystal Growth: Substructure and Recrystallization
73
6. Recrystallization of single crystals
Because of relatively low perfectness of single crystals of some semiconductors and
refractory metals, grown by EBFZM, there are some studies made to improve the
substructure of single crystals. It is well known that the growth of the semiconductor alloys
crystals is one of old problems in physics and practice of the crystal growth. The studies on
growing such crystals include the casting-recrystallizing-annealing procedures and require
careful balancing of the pseudo binary melt stoichiometry, which inherently is a quite
difficult process. Increase in recrystallization efficiency can be achieved by adjusting the
casting conditions and the suitable thermal gradient during recrystallization (Yadava et al.,
1985). This means, that the mechanism of the crystal growth is a combination of both the
chemical potential gradient and temperature gradient of zone melting processes, so the
growing processes of semiconductor alloys crystals are complicated at their practical
realization.
In comparison with the growth of semiconductor crystals, the growth of such simple metals
like molybdenum and tungsten seems to be a relatively non-problematic task. However, the
main obstacles are both the high melting temperatures and high temperature gradients
along single crystals when growing from the melt. Because of these obstacles the growth of
single crystals of refractory metals becomes the very complicated task which attracts
attention of many scientists for a half a century. Numerous studies concerning this problem
have shown that these metals, in spite of their crystallographic simplicity, require special
studies, knowledge, and equipment.
Using collected experimental information, the single crystals of molybdenum and tungsten,
free of the specific substructure, are produced by recrystallization including plastic
deformation and high-temperature annealing (Bdikin et al., 1999; Katoh et al., 1991). Plastic
deformation of the crystals has been done by rolling in the vacuum rolling machine or in the
standard mills in air. High temperature annealing is produced with the help of the heating
devices located inside the rolling machine or in the set-up for electron-beam zone melting
and growing single crystals. The comparison of structural perfection of the single crystals
grown from the melt and grown by recrystallization is done by both X-ray rocking curves
and angular scanning topography.
As a rule, the recrystallized single crystals of molybdenum and tungsten have the
substructures characterized by both the record-low dislocation density and small-angle
mosaic. It should be noted that the lower dislocation density in the single crystals can only
be achieved by recrystallization. As mentioned already before, the optimal procedure
involves deformation of the single crystals with the <111> growth axis by rolling along the
(112) plane (Bozhko et al., 2008). To monitor both the substructure and perfection of the
tungsten single crystals, the anomalous X-rays transmission has been employed as well.
6.1 Experiments on recrystallization of the tungsten single crystals
The tungsten single crystals under study are grown by the EBFZM method (Glebovsky et al.,
1986). The single crystals have the different crystallographic growth axes. The high-purity
tungsten powders of chemical purity 99.99% are used as a starting material. The as-grown
single crystals are 11-22 mm in dia and 100 mm in length. The structural studies involve the
X-ray diffraction microscopy methods, namely, angular scanning topography and rocking
Recrystallization
74
curves (Aristov et al., 1974; Bozhko et al., 2008; Brunner & Glebovsky, 2000a, 2000b; Ermolov
et al., 1999, 2002; Riedle et al., 1994, 1996).
Several deformation systems are studied, which differ by the crystallographic parameters
and strain, because two parameters such as the growth axis and the deformation direction
are most important to get the strained single crystals before high-temperature annealing.
Crystallographic systems tested are [100]/(010), [100]/(011), [110]/(110), [110]/(111)],
[111]/(110) and [111]/(112) where [growth axis]/(rolling plane). The 6-12% deformation by
rolling of the two-cant bars with fixed crystallography is found to be optimal to get the large
grains. The vacuum conditions are most suitable for the crystal deformation by rolling
because they enable one to avoid oxidation of the crystal surfaces during deformation at
high temperatures.
Fig. 15. Two-cants bar with fixed crystallography for rolling. The crystallographic
parameters are for the system [111]/(112) where [111] is the growth axis and (112) is the
rolling plane (P).
Sample Growth axis Rolling plane Strain, % Size of grains
1 [100] (010) 13.4 Grains, 8 mm
2 [100] (011) 11.6 Grains, 8 mm
3 [100] (010) 12.3 Grains, 8 mm
4 [100] (011) 11.6 Grains, 8 mm
5 [110] (110) 7.9 Grains, 15 mm
6 [110] (111) 7.5 Grains, 15 mm
7 [111] (110) 7.5 Grains, 15 mm
8 [111] (112) 6.6 Grains, 25 mm
Table 4. Parameters of the samples subjected to vacuum rolling followed by high-
temperature annealing.
Crystal Growth: Substructure and Recrystallization
75
Fig. 16. Electron-beam annealing of a single crystal in a container, a-EB gun, b-container.
The billets to be plastically deformed are produced from the cylindrical tungsten single
crystals and have a form of the two-cant bar so that crystallography has been fixed before
rolling (Fig. 15). Plastic deformation is performed in one pass at the temperature 900
0
C in
the vacuum rolling machine or, in a number of cases, in the standard rolling mill in air. In
Table 4 are listed the main data for the recrystallized single crystals. The hypercritically
strained tungsten single crystals are annealed at 2500
0
C in the EBFZM set-up by the
defocused electron beam (Fig. 16, a). In order to obtain the more uniform temperature field
in the specimens, the tungsten container is used. Its height is 40 mm, the diameter 30 mm,
the wall thickness 3 mm (Fig. 16, b). The tungsten specimen of the maximal length 30 mm
and the diameter 20 mm is installed inside the container on the tungsten holders. The
container has been arranged coaxially with the EB gun equipped with a circular cathode.
6.2 Recrystallized tungsten single crystals in comparison with ones grown from the
melt
In order to elucidate influence of recrystallization on the real substructure of tungsten single
crystals, the topograms of angular scanning of the specimens, cut from the melt-grown
single crystal before deformation, are taken. In Fig. 17, the topograms show the mosaic
substructures of two as-grown tungsten single crystals grown in identical conditions. The
size of the subgrains is about 1-2 mm, the misorientation angles between subgrains are
about 50 of an arc. On the angular scanning topograms are well seen the small-angle
boundaries, their misorientation angles make up tens of the angular minutes, for some
single crystals they can exceed 1 of an arc. Single crystals of such structural quality are not
suitable for producing, i.e., deflectors and targets using in the runs concerned with
channeling of high-energy particles beams. It is necessary to decrease substantially the
dislocation density in such crystals and to deplete them from the small-angle boundaries. In
contrast to the melt-grown single crystals, necessary structural perfection can only be
achieved by the recrystallization processes consisting of plastic deformation and high-
temperature annealing.
Recrystallization
76
Fig. 17. Angular scanning topograms of two as-grown tungsten single crystals, (110)
reflection, CuK
.
The rocking curves are recorded in the dispersion-free system using the perfect silicon single
crystal as a monochromator (Kittel, 1996). To measure the rocking curves, the angle between the
X-ray source and the detector is fixed under the first order Bragg diffraction angle. A crystal is
rotated with respect to the incoming X-ray beam, and an intensity of the diffracted beam is
measured as a function of the angle. The X-ray diffraction rocking curves are measured in three
different sites on a specimen surface. The sites differ in a rotation angle of 45
0
in order to
exclude texture effects. A half-width (FWHM) of the rocking curves is ~1
0
of an arc. Different
subgrains show up as individual peaks on the rocking curves. The angular scanned topograms
are obtained in the -2 scanning regime and used to study the position, dimensions, and
misorientation angles of subgrains. This X-ray diffraction technique is based on a principle of
the Bragg reflection (Bdikin et al., 1999; Cortenraad et al., 2001a, 2001b, 2001c, 2001d).
In Table 1 are listed main data for those single crystals. The final recrystallized specimen is
the polycrystal incorporating the distinct large grains (Fig. 18). After the recrystallization
procedure the large grains do not contain the small-angle boundaries and have the relatively
low dislocation density. The optimal procedure, involving deformation of the single crystal
with the <111> growth axis by rolling along the (112) plane, is the most suitable for
recrystallization. The ~6% deformation is found to be optimal for these crystallographic
parameters to get few single nuclei and then large grains to be grown. Higher deformation
leads to too many nuclei and as a result - to smaller grains grown. The vacuum conditions
are most suitable for rolling because they enable one to avoid oxidation of the crystal
surfaces during deformation at high temperatures. By this technique the high purity single
crystals of the low dislocation density and free of the small angle boundaries are produced.
In several cases the rolled specimens are annealed outside the container. That results in
formation of the surface damaged layer of up to 500 m thick. The middle part of such
specimens is virtually free from the small-angle boundaries and has the perfect structure
although there are many small subgrains at the periphery.
Crystal Growth: Substructure and Recrystallization
77
Fig. 18. Recrystallized tungsten polycrystal incorporating three distinct large grains.
Using a tungsten container for high-temperature annealing enables both avoiding damage
of the specimen surface layer and decrease of the temperature gradient. The specimens with
record structural perfection are obtained by using this technique. The angular scanning
topogram and corresponding rocking curve, taken from the specimen annealed in the
container are shown in Fig. 19. The angular scanning topogram shows the perfect surface of
the specimen, without any boundary or other defects (compare with the topogram for as-
grown specimens in Fig. 17). The rocking curve has only one sharp peak with the width at a
half height (FWHM) of ~50 of an arc. Correspondingly, the dislocation density is about
5x10
4
cm
-2
for this specimen.
Strong changes of the substructure of the perfect tungsten single crystal has been found
when the latter has been used as a seed crystal for growing a new tungsten crystal from
the melt by EBFZM (Glebovsky & Semenov, 1999). Before growing, the dislocation density
of the seed crystal is of about 10
4
cm
-1
, and it does not contain any subgrains. After
growing from the melt, the substructure of the seed single crystal indicates significant
deterioration - subgrains of 500 m are appeared, and they are elongated along the
growth axis with the misorientation angle of 8-10 of an arc (Fig. 20). At the growth of
single crystals of refractory metals from the melt, the main contribution to formation of
the substructure is made by the dislocations arising under influence of the thermal
stresses during the growth and cooling of the growing crystal. During growing from the
melt, dislocations may arise due to thermal stresses in the solid and under action of the
impurity concentration gradients due to lattice oversaturation with vacancies. Subgrains
are formed by fresh dislocations and walls in spite of the fact that the seed crystal does
not contain any subgrains before growing.
Recrystallization
78
Fig. 19. Angular scanning topogram (a) and rocking curve (b) of the tungsten single crystal,
annealed in the container; the surface plane (110).
Crystal Growth: Substructure and Recrystallization
79
Fig. 20. Appearance of the fresh dislocation walls in the tungsten single crystal during the
growth from the melt.
6.3 Anomalous X-ray transmission in the tungsten single crystals
A diverging X-ray beam is the beam where the angles of incidence of the X-rays on the
specimen differ for different parts of the specimen (Bdikin et al., 1999). Thus, for the given
wavelength the diffraction conditions are satisfied along a certain line on the surface of the
specimen. The classical system is used to record anomalous X-ray transmission (Borrmann
effect). A fine-focus tube with a focus measuring 50x50 m
2
is used as a source of the
diverging X-ray beam. The distance between the specimen and the photographic film is 10
cm and that between the source and the specimen is 18 cm (Fig. 21). An asymmetric
extraction geometry (the distance between the source and the specimen is not equal to that
between the specimen and the photographic film) is used to eliminate focusing of the
diffracted rays over the radiation spectrum. In the case of dynamic diffraction (perfect
crystal) at points, where the diffraction conditions are satisfied, amplification of intensity is
observed in the transmitted beam. In kinematic approximation (crystal containing defects),
intensity is suppressed. The fact that transmitted radiation is recorded, and radiation
intensity is the same in the diffracted and forward transmitted beams, together with
observation of amplification of characteristic CuK
1,2
radiation intensity in these directions
indicate that diffraction is of dynamic nature. Observation of splitting of these beams into a
CuK
1,2
doublet (Fig.22) and a shift as the specimen rotates indicate that these beams are of
diffraction origin and are not the topological characteristic of the specimen.
Diffractometering of the waves which passed in the directions R and T also confirms that
intensities of the diffracted and transmitted beams are the same.
Recrystallization
80
Fig. 21. Scheme for recording the anomalous transmission of X-rays. R-reflection, T-
transmission.
Fig. 22. Anomalous X-rays transmission in a perfect W single crystal; lines are from
transmitted beams.
Calculations using the kinematic theory show that for Cu-radiation and the specimen
thickness of 0.3 mm (t ~ 98) absorption is 10
43
. With the available radiation sources (no
more than 10
7
-10
8
pulse/s) it is impossible to obtain recordable transmission intensity. For
Mo-radiation, absorption is lower than that for Cu-radiation. The linear absorption
coefficient of the 0.3 mm thick tungsten single crystal is calculated to be t = 53.4 on the Mo-
radiation characteristics.
Crystal Growth: Substructure and Recrystallization
81
If the absorption coefficient is conceived as =
i
t + y
i
, where
i
is a coefficient of
interference absorption and y
i
is responsible for renormalization of the atomic amplitude of
scattering, then for the perfect tungsten single crystal can be obtained
i
= 369.2, y
i
= -6.63.
The measurements of the absorption coefficient of a diffracted Mo-radiation yield = 7.3.
Under assumption that y
i
is independent of the defects concentration one obtains
i
=464.3
cm
-1
, i.e., in the real specimen the interference absorption coefficient is by 25% greater than it
has to be in the perfect crystal.
The effects associated with dynamic X-ray propagation can only be observed in crystals for
which the distance between the dislocations does not exceed an extinction length L. The
calculations for (110) CuK
}
(2)
where l, and v are the series expansion orders and
l
C
v
is the corresponding expansion
coefficient. For a random texture, J=1.0 and for an ideal texture of single orientation, J.
Generally, heavier plastic deformation leads to sharper texture. In addition, the TOF neutron
spectrum measurement covers a wide range of lattice plane spacings which is partially
equivalent to measuring one reflection over a wide range of sample orientations. Therefore,
the texture indexes calculated from the TOF neutron spectra by the GSAS software package
(Larson, 2004) can be employed to compare the plastic strains of austenite and ferrite.
As shown in Fig.14, the texture indexes of the ferrite phase are evidently larger than those of
austenite in both the axial and radial directions while the difference between the two directions
is related to the details of the crystallographic orientation distribution. This suggests that the
plastic deformation occurs preferentially in the ferrite matrix during warm compression.
The different peak shifts in Section 4.2 & Section 4.3 reveal that the austenite grains are
harder than the ferrite grains. The texture index of the ferrite matrix is larger than that of
austenite, suggesting that the plastic deformation occurs preferentially in the ferrite matrix.
According to the related recrystallization literature (Doherty, 1997), the recrystallized ferrite
grains nucleate preferentially in the heterogeneously deformed regions near large hard
particles. Therefore, the existence of carbon-enriched austenite is able to accelerate the
dynamic recrystallization in the ferrite matrix.
Recrystallization
104
Fig. 14. Change in texture indexes of austenite and ferrite during warm compression: (a)
obtained from the axial direction; (b) obtained from the radial direction. (Xu, 2008b)
The TOF neutron diffraction experiment helped us to understand the austenite precipitation
and dynamic recrystallization behaviors of the 17Ni-0.2C martensite steel during isothermal
warm compression after isothermal holding at 773K, and the main results were as follows:
(1) the warm deformation at 773K accelerated the austenite precipitation in 17Ni-0.2C
martensite steel. Splitting of the austenite (111) peak was found to occur and then disappear
during warm compression. The splitting is ascribed to the different carbon concentrations in
the dynamically precipitated austenite and the pre-existing austenite, and the disappearance
of splitting is related to carbon homogenization at larger strain. (2) The austenite in the
17Ni-0.2C martensite steel is harder than the ferrite matrix at 773K. Heterogeneous
deformation occurs preferentially in the ferrite matrix, leading to the acceleration of
dynamic recrystallization.
5. Summary and future works
The low-carbon martensite steel and the high-nickel martensite steels have been warm
compressed with and without prior tempering treatment to realize the microstructure
refinement based on dynamic recrystallization at lower Zener-Hollomon parameter Z.
Because of the lower critical strain for fully recrystallization compared with the warm
deformation of ferrite or pearlite/ferrite, the warm deformatioin of martensite may be
employed to the production of future ultrafine grained multiphase steels. The carbon
addition promotes the formation of hard second phase particles such as cementite and
austenite, which accelerates the ferrite recrystallization through the formation of local high
strain regions near the hard particles during warm deformation.
Neutron diffraction has been applied as a powerful tool to investigate the microstructure
evolutions of bulk materials during tensile/compressive deformation, heating/cooling and
under other specific environmental conditions
(te Velthuis, 1998; Tomota, 2005; Xu, 2006a;
Xu, 2006b). Recently, TOF (hkl) multiple reflection spectra obtained by neutron diffraction
have been analyzed to evaluate the crystallographic textures during forward and reverse
diffusional phase transformations
(Wenk, 2007) and the preferred orientations of ferrite and
austenite in a 0.2C-2Mn steel before and after hot compression (Xu, 2009). The in situ
Recrystallization Behavior During Warm Compression of Martensite Steels
105
microstructure and texture evolution during thermomechical controlled process will be
studied further in order to well optimize the multiphase microstructure (Xu, 2012).
6. Acknowledgments
The authors thank Dr. Y. Adachi at National Institute for Materials Science, Japan for his
support on the TEM microstructure observation. They also appreciate Dr. E.C. Oliver at ISIS
Facility, Rutherford Appleton Laboratory, United Kingdom for his support on the neutron
diffraction.
7. References
Ameyama, K.; Matsumura, N. & Tokizane, M. (1988). Ultrafine Austenite Grains Obtained
by Thermomechanical Processing in Low and Medium Carbon Steels. Journal of the
Japan Society for Heat Treatment, Vol.28, no.4, pp.233-240, ISSN 0288-0490.
Bao, Y.Z.; Adachi, Y.; Toomine, Y.; Suzuki, T.; Xu, P.G. & Tomota, Y. (2005a). Dynamic
Recrystallization Behavior in Martensite in 18Ni, 17Ni-0.2C and SM490 Steels.
Tetsu-to-Hagan (Journal of the Iron and Steel Institute of Japan), Vol.91, pp. 602-
608, ISSN 0021-1575.
Bao, Y.Z.; Adachi, Y.; Toomine, Y.; Xu, P.G.; Suzuki, T. & Tomota, Y. (2005b). Dynamic
Recrystallization by Rapid Heating Followed by Compression for a 17Ni-0.2C
Martensite Steel. Scripta Materialia, Vol.53, pp. 1471-1476, ISSN 1359-6462.
Bunge, H.J. (1982) Texture Analysis in Materials Science. Butterworth & Co., ISBN 0-408-10642-
5, London, pp.88-98.
Chen, S.C.; Tomota, Y.; Shiota, Y.; Toomine, Y. & Kamiyama, T. (2006). Measurements of
Volume Fraction and Carbon Concentration of the Retained Austenite by Neutron
Diffraction. Tetsu-to-Hagan (Journal of the Iron and Steel Institute of Japan), Vol.92,
pp.557-561, ISSN 0021-1575.
Doherty, R.D.; Hughes, D.A.; Humphreys, F.J.; Jonas, J.J.; Juul Jensen, D.; Kassner, M.E.;
King, W.E.; McNelley, T.R.; McQueen H.J. & Rollett, A.D. (1997). Current issues in
recrystallization: a review. Materials Science and Engineering A, Vol.238, pp.219-274,
ISSN 0921-5093.
Dong, H. & Sun, X.J. (2006). Deformation induced ferrite transformation in low carbon
steels. Current Opinion in Solid State and Materials Science, Vol.9, pp.269-276, ISSN
1359-0286.
Enomoto, M. & Furubayashi, E. (1977). A Crystallographic Study of Austenite Formation
from Fe-Ni Martensite during Heating in Alpha-Gamma Region. Transactions of the
Japan Institute of Metals, Vol.18, pp.817-824, ISSN 0021-4434.
Furuhara, T.; Yamaguchi, T.; Furimoto, S. & Maki, T. (2007). Formation of
Ferrrite+Cementite microduplex structure by warm deformation in high carbon
steels. Material Science Forum, Vol.539-543, pp.155-160, ISSN 0255-5476.
Glovre, G. & Selllers, C.M. (1973). Recovery and recrystallization during high temperature
deformation of o-iron. Metallurgical Transactions, Vol.4, pp. 765-775, ISSN 0026-
086X.
Recrystallization
106
Hayashi, T.; Torizuka, S.; Mitsui, T.; Tsuzaki, K. & Nagai, K. (1999). Creation of low-carbon
steel bars with fully fine ferrite grain structure through warm grooved rolling.
CAMP-ISIJ (Current Advances in Materials and Processes), Vol.12, pp.385-388,
ISSN 1882-8922.
Hayashi, T. & Nagai, K. (2002). Inprovement of strength-ductility balance for low carbon
ultrafine-grained steel through strain hardening design. Transactions of the Japan
Society of Mechanical Engineers. A, Vol.68, pp.1553-1558, ISSN 0387-5008.
Hutchings, M.T.; Withers, P.J.; Holden, T.M. & Lorentzen, T. (2005). Introduction to the
Characterization of Residual Stress by Neutron Diffraction, ISBN 0-415-31000-8, Taylor
& Francis, New York, pp.230-238.
Larson, A.C. & Von Dreele, R.B. (2004). General Structure Analysis System (GSAS), Los
Alamos National Laboratory Report. LAUR 86-748, pp.147-148.
Li, J.H.; Xu, P.G.; Tomota, Y. & Adachi, Y. (2008) Dynamic Recrystallization Behavior in a
Low-carbon Martensite Steel by Warm Compression. ISIJ International, Vol.48, no.7,
pp.1008-1013, ISSN 1485-1664.
Maki, T.; Okaguchi, S. & Tamura, I. (1982). Dynamic recrystallization in ferritic stainless
steel. Strength of metals and alloys (ICSMA 6) : Proceedings of the 6th International
Conference, pp.529-534., ISBN 0080293255. Melbourne, Australia, 16-20 August
1982.
Maki, T.; Furuhara, T.; & Tsuzaki, K. (2001). Microstructure Development by
Thermomechanical Processing in Duplex Stainless Steel. ISIJ International, Vol..41.
pp. 571-579. ISSN 1485-1664.
Miller, R.L. (1972). Ultrafine-grained microstructures and mechanical properties of alloy
steels. Metallurgical Transactions, Vol.3, pp.905-912, ISSN 0026-086X.
Moriyama, M.; Takaki, S. & Kawagoishi, N. (2001). Influence of Reversion Austenite on
Fatigue Property of 350 ksi Grade 18Ni Maraging Steel. Journal of the Japan Society
for Heat Treatment, Vol.41. pp.266-271, ISSN 0288-0490.
Najafi-Zadeh, A.; Jonas, J.J. & Yue, S. (1992). Grain refinement by dynamic recrystallization
during the simulated warm-rolling of interstitial free steels. Metallurgical
Transactions A, Vol.23, pp.2607-2617. ISSN 0360-2133.
Ohmori, A. ; Torizuka, S. ; Nagai, K. ; Yamada K. & Kogo, Y. (2002). Evolution of ultrafine-
grained structure through large strain-high Z deformation in a low carbon steel.
Tetsu-to-Hagan (Journal of the Iron and Steel Institute of Japan), Vol.88, no. 12,
pp.857-864, ISSN 0021-1575.
Ohmori, A.; Torizuka, S.; Nagai, K.; Yamada, K. & Kogo, Y. (2004). Effect of Deformation
Temperature and Strain Rate on Evolution of Ultrafine Grained Structure through
Single-Pass Large-Strain Warm Deformation in a Low Carbon Steel. Materials
Transactions, Vol.45, pp.2224-2231, ISSN 1345-9678.
Poorganji, B.; Miyamoto, G.; Maki, T. & Furuhara, T. (2008). Formation of ultrafine grained
ferrite by warm deformation of lath martensite in low-alloy steels with different
carbon content. Scripta Materialia, Vol.59, pp.279-281, ISSN 1359-6462.
Reed, R.C. & Root, J.H. (1998). Determination of the Temperature Dependence of the Lattice
Parameters of Cementite by Neutron Diffraction. Scripta Materialia, Vol.38, pp.95-
99. ISSN 1359-6462.
Recrystallization Behavior During Warm Compression of Martensite Steels
107
Sawada, K.; Ohba, T.; Kushima, H. and Kimura, K. (2005). Effect of microstructure on elastic
property at high temperatures in ferritic heat resistant steels. Materials Science and
Engineering A, Vol.394, pp.36-42, ISSN 0921-5093.
te Velthuis, S.G.E.; Root, J.H.; Sietsma, J.; Rekveldt, M.T. & van der Zwaag, S. (1998). The
ferrite and austenite lattice parameters of Fe-Co and Fe-Cu binary alloys as a
function of temperature. Acta Materialia. Vol.46, pp.5223-5228, ISSN 1359-6454.
Tomota, Y.; Suzuki, T.; Kanie, A.; Shiota, Y.; Uno, M.; Moriai, A.; Minakawa, N. & Morii, Y.
(2005). In situ neutron diffraction of heavily drawn steel wires with ultra-high
strength under tensile loading. Acta Materialia. Vol.53, pp.463-467, ISSN 1359-6454.
Tomota, Y.; Narui, A. & Tsuchida, N. (2008). Tensile behavior of fine-grained steels, ISIJ
International, Vol.48, pp.1107-1113, ISSN 1485-1664.
Torizuka, S. (2005) Production of ultrafine-grained steel bar and plate by high Z-large strain
deformation in ferrite region. Ferrum (Bulletin of The Iron and Steel Institute of
Japan), Vol.10, no.3, pp.188-195, ISSN 1341-688X.
Tsuji, N.; Matsubara, Y.; Saito, Y. & Maki, T. (1998). Occurance of Dynamic Recrystallization
in ferritic Iron. Journal of the Japan Institute of Metals, Vol.62, pp.967-976. ISSN 0021-
4876.
Tsuji, N.; Okuno, S.; Koizumi, Y. & Minamino, Y. (2004). Toughness of Ultrafine Grained
Ferritic Steels Fabricated by ARB and Annealing Process. Materials Transactions,
Vol. 45, no.7, pp.2272-2281, ISSN 1345-9678.
Ueji, R.; Tsuji, N.; Minamino, Y.; & Koizumi, Y. (2002). Ultragrain refinement of plain low
carbon steel by cold-rolling and annealing of martensite. Acta Materialia, Vol.50,
pp.4177-4189, ISSN 1359-6454.
Wenk, H.R.; Huensche, I. & Kestens, L. (2007). In-Situ Observation of Texture Changes
during Phase Transformations in Ultra-Low-Carbon Steel. Metallurgical and
Materials Transactions A, Vol.38, pp.261-267, ISSN 1073-5623.
Xu, P.G.; Tomota, Y.; Lukas, P.; Muransky, O. & Adachi, Y. (2006a). Austenite-to-ferrite
transformation in low alloy steels during thermomechanically controlled process
studied by in situ neutron diffraction. Materials Science and Engineering A, Vol.435,
pp.46-53, ISSN 0921-5093.
Xu, P.G. and Tomota, Y. (2006b). Progress in materials characterization technique based on
in situ neutron diffraction. Acta Metallurgica Sinica, Vol.42, pp.681-688, ISSN 0412-
1961.
Xu, P.G.; Li, J.H.; Tomota, Y. and Adachi, Y. (2007). Effect of Carbon Addition on Ultrafine
Grained Microstructure Formation by Warm Compression for Fe-18Ni Alloys,
Materials Science Forum, Vol.558-559, pp.601-606, ISSN 0255-5476.
Xu, P.G.; Li, J.H.; Tomota, Y. & Adachi, Y. (2008a). Effects of Volume Fraction and Carbon
Concentration of Austenite on Formation of Ultrafine Grained Ferrite/Austenite
Duplex Microstructure by Warm Compression, ISIJ International, Vol.48, no.11,
pp.1609-1617, ISSN 1485-1664.
Xu, P.G.; Tomota, Y. & Oliver, E.C. (2008b). Dynamic Recrystallization and Dynamic
Precipitation Behaviors of a 17Ni-0.2C Martensite Steel Studied by In Situ Neutron
Diffraction, ISIJ International, Vol.48, no.11, pp.1618-1625. ISSN 1485-1664.
Recrystallization
108
Xu, P.G.; Tomota,Y.; Suzuki, T.; Yonemura, M. & Oliver, E.C. (2009). In Situ TOF Neutron
Diffraction for Isothermal Ferrite Transformation during Thermomechanically
Controlled Process of Low Alloy Steel. Netsu Shori (Journal of the Japan Society for
Heat Treatment), Vol.49. special issue, pp.470-473. ISSN 0288-0490.
Xu, P.G.; Tomota, Y.; Vogel, S.C.; Suzuki, T.; Yonemura, M. & Kamiyama, T. (2012)
Transformation Strain and Texture Evolution during Diffusional Phase
Transformation of Low Alloy Steels Studied by Neutron Diffraction, Reviews on
Advanced Materials Science. Vol.33. (in press). ISSN 1605-8127.
6
The Deformability and Microstructural Aspects
of Recrystallization Process in Hot-Deformed
Fe-Ni Superalloy
Kazimierz J. Ducki
Silesian University of Technology
Poland
1. Introduction
The behaviour of metals and alloys during hot plastic working has a complex nature and it
varies with the changing of such process parameters as (Zhou et al., 1994): deformation,
strain rate and temperature. The high-temperature plastic deformation is coupled with
dynamic recovery and recrystallization processes which influencing the structure and
properties of alloys. One of crucial issues is finding the relationship between the hot plastic
deformation process parameters, microstructure and properties. Since the sixties of the last
century, theoretical and experimental investigations have been carried out to find those
interdependence for steels and nickel alloys.
In the recent years, the constitutive equations describing hot plastic deformation processes
have started to take into account the so-called internal variables determining the material
condition. These variables include substructural parameters such as (Hansen, 1998; Sellars,
1998): grain size, grain shape, recrystalized volume fraction, dislocation density, subgrain
size, subgrain misorientation angle and stacking fault energy (SFE). Determination of the
above-mentioned parameters of a deformed material structure description requires the
application of analytical methods primarily based on quantitative metallography and
transmission electron microscopy (TEM). Taking those substructure parameters into
consideration in calculations should enable the correct modelling of structural phenomena
during hot plastic deformation and enhance the technological processes control for the
purpose of obtaining the assumed structures of required properties (McQeen et al., 2002).
The Fe-Ni superalloys precipitation hardened by intermetallic phase of ' - Ni
3
(Al,Ti) type
are one of the groups of construction materials intended for operation in cryogenic and
elevated temperatures. These alloys are difficult to deform and are characterized by high
values of yield stress at a high temperature. High deformation resistance of Fe-Ni alloys is
caused by a complex phase composition, high activation energy of the hot plastic
deformation process and a low rate of dynamic recrystallization. When choosing the
conditions for hot plastic working of Fe-Ni alloys, the following factors should be
considered (Bywater et al., 1976; Kohno et al., 1981; Ducki et al., 2006): the matrix grain size,
plastic deformation parameters and the course of the recrystallization process. The grain
size is of particular importance. Grain refining leads to an increased rate of recovery and
Recrystallization
110
dynamic recrystallization and to a smaller diameter of recrystallized grains. This is
important, for the grain refinement in Fe-Ni superalloys has an advantageous influence on
increasing their yield point and fatigue strength (Koul et al., 1994; Hrkegrd et al., 1998).
In the presented study, research has been undertaken on the influence of the initial
microstructure of austenite and the parameters of hot plastic working on deformability,
grain and subgrain size, and dislocation density in a high-temperature creep resisting Fe-Ni
alloy. It is assumed that the results obtained will be used for optimizing hot plastic working
processes and forecasting the microstructure and functional properties of products made of
Fe-Ni superalloys.
2. Material and methodology
The examinations were performed on rolled bars, 16 mm in diameter, of an austenitic A-286
type alloy. The chemical composition is given in Tab. 1.
Content of an element [wt. %]
C Si Mn P S Cr Ni Mo V W Ti Al Co B N Fe
0.05 0.56 1.5 0.026 0.016 14.3 24.5 1.35 0.42 0.10 1.88 0.16 0.08 0.007 0.0062 55.3
Table 1. Chemical composition of the investigated Fe-Ni superalloy
In order to model the conditions of alloy heating prior to plastic processing, the
investigations were carried out on samples after initial soaking at high temperatures.
Sections of rolled bars, which the samples for investigations were made of, were subjected to
two variants of preheating, i.e. 1100C/2h and 1150C/2h with subsequent cooling in water.
Heat treatment of this type corresponds to the soaking parameters of the investigated
superalloy before hot plastic processing (Kohno et al., 1981).
The research on the alloy deformability was performed in a hot torsion test on a Setaram
torsion plastometer 7 MNG. The plastometric tests were performed every 50C in a
temperature range of 9001150C, with a constant holding time of 10 minutes at the defined
temperature. Solid cylindrical specimens ( 6.0 50 mm) were twisted at a rotational speed
of 50 and 500 rpm, which corresponds to the strain rate of 0.1 and 1.0 s
-1
, respectively. To
freeze the structure, the specimens after deformation until failure were directly rapid cooled
in water (Fig. 1).
The data obtained in the plastometric torsion test were entered in an Excel spreadsheet in
the form of columns containing the recorded values. Processing of the measured data by
means of filtration, cutting, shrinking and planishing was conducted using the Matlab 6
program. A correction of the torque moment, due to diversified rotational speed values and
increase of the sample temperature during torsion, was calculated by the method of joint
action of speed and temperature from the following relations (Hadasik, 2005):
''' ''' M M M = + A (1)
''' ( , , ) ( , , )
r
M M N N T M N N T T A = + A
(2)
The Deformability and Microstructural Aspects of
Recrystallization Process in Hot-Deformed Fe-Ni Superalloy
111
''' exp( ) exp exp( ) exp
E E
D D
B B
T T T
F F
M A N C N N A N C N N
T T T
+ +
+A
| | | |
A =
| |
+ A
\ . \ .
(3)
where: M recorded torque moment [Nm]; M''' corrected torque moment value [Nm];
AM''' torque moment correction taking account of a joint action of speed and temperature
[Nm]; N number of sample torsion, N
recorded
rotational speed [rpm]; T deformation temperature [C]; AT temperature increment
during torsion [C]; A, B, C, D, E, F material constans.
Fig. 1. Scheme of plastometric examination of the Fe-Ni alloy
The corrected data constituted a basis for the determination of equivalent deformation c as a
function of the number of the samples rotations during torsion (Hadasik, 2005):
2
3
arcsin
RN
h
L
t
c
| |
= |
|
\ .
(4)
where: R equivalent radius corresponding to 2/3 of the outer radius R of the sample, L
measured sample length.
Yield stress o
p
was determined according to relation (5) taking account of the corrected
torque moment M''', sample radius R, parameters m, p and axial force F
o
(Hadasik, 2005):
( )
0 5
2
2
3 2
3
3
2
,
'''
p
M F
p m
R R
o
t t
(
| |
| |
(
= + + + |
|
|
(
\ .
\ .
(5)
where: p parameter reflecting stress sensitivity to deformation size; m parameter
reflecting stress sensitivity to deformation rate.
On the flow curves determined, the following parameters characterising plastic properties of
the alloy in the torsion test were defined:
Recrystallization
112
- o
pp
maximum yield stress on the flow curve;
- c
p
deformation corresponding to the maximum yield stress;
- o
f
stress at which the sample is subject to failure;
- c
f
deformation at which the sample is subject to failure, the so-called threshold
deformation.
Relations between the yield stress and alloy deformation, and the deformation conditions
were described using the Zener-Hollomon parameter Z (Zener et al., 1944):
( )
exp sinh
n
pp
Q
Z A
RT
c oo
| |
(
= =
|
\ .
(6)
where: c
+
= [m
-2
] (12)
where: x a coefficient which defines the fraction of invisible dislocations with Burgers
vectors a/2<111> for the A1 structure: x = 2 for image of dislocations observed in (111) reflex,
x = 1.5 for image of dislocations observed in (200) reflex, x = 1.5 for image of dislocations
observed in (220) reflex; l
1(2)
the total length of the horizontal (vertical) lattice lines; n
1(2)
the number of intersections of the horizontal (vertical) lattice lines with dislocations; t the
thickness of the foil.
Recrystallization
114
The thickness of the foil in the investigated areas can be approximately calculated following
the formula (Head et al., 1973):
t = n
hkl
(13)
where: n a number of extinction lines;
hkl
a value of extinction.
The values of
hkl
given by (Head et al., 1973) must be considered as a rough estimation of
the actual value of extinction for the investigated material.
3. Results and discusion
3.1 Initial microstructure of the alloy
The application of two variants of initial solution heat treatment simulating the heating
conditions allowed diversifying significantly the initial microstructure of the Fe-Ni alloy
before hot plastic deformation. The alloy in its initial state differed primarily in the average
grain size. After solution heat treatment under the conditions of 1100C/2h/w., in the alloy
microstructure presence was found of twinned austenite with medium-size grain
( = 2120 m
2
) with a small amount (ca. 0.3 wt.%) of insoluble particles (Fig. 3a). The
increasing of the solution heat treatment temperature to 1150C at an analogous soaking
time resulted in an increase of the austenite grain ( = 6296 m
2
) and a reduction in the
quantity and size of undissolved primary particles (Fig. 3b). For the analyzed variants of
solution heat treatment, the microstructure of the samples was characterized by equiaxial
grains, as evidenced by the elongation coefficient approximate to 1 and the dimensionless
shape factor of approximately 0.6.
Presence of titanium compounds, such as TiC carbide, TiN
0.3
nitride, TiC
0.3
N
0.7
carbonitride,
Ti
4
C
2
S
2
carbosulfide and Lavesa Ni
2
Si phase and boride MoB was disclosed in the phase
composition of the undissolved particles (Ducki, 2010).
a) b)
Fig. 3. Diversified microstructure of alloy after solution heat treatment at: a) 1100C/2 h/w.,
= 2120 m
2
; b) 1150C/2 h/w., = 6296 m
2
The Deformability and Microstructural Aspects of
Recrystallization Process in Hot-Deformed Fe-Ni Superalloy
115
3.2 Deformability of the alloy
Knowledge of the phenomena occurring during hot deformation of materials enables
selecting the correct conditions for plastic working and shaping their material
characteristics. The results of the plastometric investigations, in the form of the calculated
alloy flow curves at temperatures of 9001150C for two options of initial soaking are shown
in Fig. 4 and 5. The curves obtained for the option of initial soaking at 1100C/2h and strain
rate 0.1 s
-1
have a shape characteristic of a material in which dynamic recovery and
recrystallization take place (Fig. 4). High deformation values were obtained for the alloy in a
wide range of torsion temperatures, i.e. 9501100C. An increase of strain rate to 1.0 s
-1
results in a significant increase of yield stress values and a distinct decrease of the alloy
deformability at all temperatures analysed. This phenomenon can be explained by a higher
alloy consolidation rate as well as too slow removal of the reinforcement as a result of
dynamic recovery and recrystallization.
Fig. 4. The effect of deformation temperature on the flow stress of Fe-Ni alloy after initial
soaking at 1100C/2 h. Strain rate: 0.1 s
-1
and 1.0 s
-1
Recrystallization
116
Fig. 5. The effect of deformation temperature on the flow stress of Fe-Ni alloy after initial
soaking at 1150C/2h. Strain rate: 0.1 s
-1
and 1.0 s
-1
An increase of the initial soaking temperature to 1150C/2h significantly reduces the alloy
deformability for the two strain rates, both at low and high deformation temperatures (Fig.
5). In this case, fairly high deformation values were obtained for the alloy in a narrow range
of torsion temperatures, i.e. 10001050C. Such behavior of the alloy may be explained by a
larger growth of austenite grains at this soaking temperature and, consequently, lower
recovery and dynamic recrystallization rates.
The values determined for the maximum yield stress o
pp
, maximum deformation c
p
, stress
until failure o
f
and threshold deformation c
f
depending on the temperature and strain rate
are presented in Figs. 6-9. For the option of initial soaking at 1100C/2h and torsion speed
of 0.1 s
-1
, the alloy under discussion shows a contiuous drop of o
pp
from values 277 MPa at
a temperature of 900C to the value of 81 MPa at 1150C (Fig. 6). The threshold
deformation c
f
rises initially together with the torsion temperature, reaching the
maximum of (3.19/3.14) at 10001050C, and then falls (Fig. 8). An increase of the strain
rate to 1.0 s
-1
results in an increase of o
pp
to maximum values of 367 MPa at the
temperature of 900C (Fig. 6) and a decrease of the threshold deformation to the
maximum of 2.47/2.34 at 10001050C (Fig. 8).
The Deformability and Microstructural Aspects of
Recrystallization Process in Hot-Deformed Fe-Ni Superalloy
117
277
216
158
128
97
81
367
309
248
189
165
125
0
50
100
150
200
250
300
350
400
900 950 1000 1050 1100 1150
Deformation temperature [C]
M
a
x
i
m
u
m
y
i
e
l
d
s
t
r
e
s
s
o
p
p
[
M
P
a
]
0.1 s
1.0 s
Strain rate
-1
-1
Soaking parameters: 1100C/2h
Fig. 6. The effect of deformation conditions on maximum flow stress. Initial alloy soaking:
1100C/2 h
293
220
168
124
99
82
370
305
247
194
151
122
0
50
100
150
200
250
300
350
400
900 950 1000 1050 1100 1150
Deformation temperature [C]
M
a
x
i
m
u
m
y
i
e
l
d
s
t
r
e
s
s
o
p
p
[
M
P
a
]
0.1 s
1.0 s
Strain rate
-1
-1
Soaking parameters: 1150C/2h
Fig. 7. The effect of deformation conditions on maximum flow stress. Initial alloy soaking:
1150C/2 h
2,05
2,57
3,19
3,14
2,73
0,78
1,26
1,45
2,47
2,34
1,93
0,36
0
0,5
1
1,5
2
2,5
3
3,5
4
900 950 1000 1050 1100 1150
Deformation temperature [C]
T
h
r
e
s
h
o
l
d
d
e
f
o
r
m
a
t
i
o
n
c
f
0.1 s
1.0 s
Strain rate
-1
-1
Soaking parameters: 1100C/2h
Fig. 8. The effect of deformation conditions on threshold deformation of the alloy. Initial
soaking: 1100C/2 h
Recrystallization
118
1,06
1,62
2,92 2,93
1,40
0,72
1,24
1,00
2,89
1,75
1,08
0,36
0
0,5
1
1,5
2
2,5
3
3,5
4
900 950 1000 1050 1100 1150
Deformation temperature [C]
T
h
r
e
s
h
o
l
d
d
e
f
o
r
m
a
t
i
o
n
c
f
0.1 s
1.0 s
Strain rate
-1
-1
Soaking parameters: 1150C/2h
Fig. 9. The effect of deformation conditions on threshold deformation of the alloy. Initial
soaking: 1150C/2 h
An increase of the alloy initial soaking temperature to 1150C/2h at a strain rate of 0.1 s
-1
results in a slight increase of o
pp
to maximum values of 293 MPa at 900C (Fig. 7) and
decrease of c
f
to the maximum of 2.922.93 in the range of 10001050C (Fig. 9). An increase
of the torsion speed to 1.0 s
-1
results in further increase of the o
pp
value to maximum values
of 370 MPa at 900C (Fig. 7), and decrease of c
f
to the maximum values of 2.89/1.75 at the
temperature of 10001050C (Fig. 9).
The activation energy of the hot plastic deformation process Q was calculated by the means
of a computer programme ENERGY 3.0 (Schindler et al., 1998). The activation energy, Q,
necessary to initiate dynamic recrystallization in the Fe-Ni alloy, was determined on the
basis of the linear dependencies presented in Fig. 10.
Fig. 10. The plot for determination of the activation energy for hot plastic deformation of the
Fe-Ni alloy. Initial alloy soaking: 1100C/2 h and 1150C/2 h
The Deformability and Microstructural Aspects of
Recrystallization Process in Hot-Deformed Fe-Ni Superalloy
119
The activation energy, Q, of hot plastic deformation for the Fe-Ni alloy depends on the
temperature of initial soaking and equals as follows:
- Q = 441.8 [kJ/mol] for initial alloy soaking 1100C/2 h;
- Q = 518.7 [kJ/mol] for initial alloy soaking 1150C/2 h.
The higher value of the activation energy, Q, of hot plastic deformation for the alloy after
initial soaking at 1150C/2h can be justified by higher values of the maximum yield stress
o
pp
, a larger growth of the initial austenite grain and a higher degree of matrix saturation
with alloying elements.
The dependencies between maximum yield stress o
pp
and Zener Hollomon Z parameter are
presented in Fig. 11. For both options of initial soaking, a power dependence (R
2
= 0,98) of
the alloy yield stress was obtained as a function of the Z parameter. So determined function
dependencies between the maximum yield stress o
pp
and the Z parameter had a form of
power function:
- for the alloy after initial soaking 1100C/2 h:
o
pp
= 0.43 Z
0.151
[MPa] (14)
- for the alloy after initial soaking 1150C/2 h:
o
pp
= 0.34 Z
0.133
[MPa] (15)
Higher values of the Z parameter for the alloy after initial soaking at 1150C/2h result from
higher values of the plastic deformation activation energy Q.
0
50
100
150
200
250
300
350
400
32 34 36 38 40 42 44 46 48 50 52 54
ln Z
M
a
k
s
i
m
u
m
y
i
e
l
d
s
t
r
e
s
s
o
p
p
[
M
P
a
]
Soaking parameters
1150C/2h
Soaking parameters
1100C/2h
R
2
=0.9852
R
2
=0.9865
Fig. 11. Dependence of the maximum flow stress o
pp
on the Zener-Hollomon parameter Z.
Initial alloy soaking: 1100C/2 h and 1150C/2 h
Recrystallization
120
3.3 Microstructure of hot-deformed alloy
The recovery and dynamic recrystallization which occur in the Fe-Ni alloy during hot plastic
deformation affect the size of the austenite grain. The results of investigations of the alloy
microstructure after initial soaking at 1100C/2h and deformation in a temperature range
9001150C and a strain rate of 0.1 and 1.0 s
-1
are presented in Figs. 12a-d.
a) b)
c) d)
Fig. 12. The microstructure of the alloy after plastic deformation at: a) 900C/1.0 s
-1
,
b) 950C/0.1 s
-1
, c) 1100C/1.0 s
-1
, d) 1150C/0.1 s
-1
. Initial soaking: 1100C/2h
After deformation at 900C for both of the torsion rates applied, the alloy structure is not
completely recrystallized, which is indicated by the presence of primary elongated grains
and fine recrystallized grains (Fig. 12a). At a torsion temperature in the range of 9501100C,
the alloy structure consisted of dynamically recrystallized grains (Fig. 12b and 12c). With an
increasing deformation temperature, a gradual growth of the recrystallized grains was
observed. The recrystallized grains were characterized by a deformed grain boundary line,
which indicates an extensive cumulated deformation in the specimens. At the highest
torsion temperature, 1150C, some elongated grains of dynamically recrystallized austenite
are observed in the alloy structure (Fig. 12d).
An increase of the initial soaking temperature of the alloy to 1150C/2h results in increasing
the initial austenite grain size and decreasing the kinetics of dynamic recovery and
recrystallization for both of the analyzed strain rates (Figs. 13a-d). After deformation at
900C/0.1 s
-1
and 900950C/1.0 s
-1
, the alloy microstructure was not completely
recrystallized and it was composed of deformed primary grains and dynamically
recrystallized grains of small sizes (Fig. 13a and 13b). The new, fine recrystallized grains
nucleated at primary grain boundaries, creating the so-called "necklace". Within the
deformation temperature range from 9501000 to 1100C, the alloy structure is fine-grained
The Deformability and Microstructural Aspects of
Recrystallization Process in Hot-Deformed Fe-Ni Superalloy
121
and completely dynamically recrystallized (Fig. 13c). The highest deformation temperature
of 1150C induced deformation localization, as evidenced by the elongated dynamically
recrystallized grains of varying sizes (Fig. 13d).
a) b)
c) d)
Fig. 13. The microstructure of the alloy after plastic deformation at: a) 900C/1.0 s
-1
,
b) 950C/0.1 s
-1
, c) 1050C/1.0 s
-1
, d) 1150C/0.1 s
-1
. Initial soaking: 1150C/2h
The results of a quantitative evaluation of the structure after initial soaking and
deformation until failure in a temperature range of 900-1150C and a strain rate of 0.1 and
1.0 s
-1
are presented in Figs. 14-17. In the structure of the alloy after initial soaking
1100C/2h and deformation in the investigated temperature range at a strain rate 0.1 s
-1
,
monotonous growth of the grain average area is observed from a value 16 m
2
at 900C
to 198 m
2
at 1150C (Fig. 14). Up to the deformation temperature of 1100C, the
dynamically recrystallized grains are approximately equiaxial (o = 0.991.12), whereas at
the highest torsion temperature, 1150C, they are elongated (o = 1.31) (Fig. 15). An
increase of the strain rate to 1.0 s
-1
induces a certain reduction of the recrystallized grain
size.
Also, in this case, in the investigated range of deformation temperatures, a monotonous
growth of the grain average area was observed, from 12 m
2
at 900C to 87 m
2
at 1100C
(Fig. 14). In the analyzed range of torsion temperatures of 9001100C, the dynamically
recrystallized grains are approximately equiaxial (o = 1.001.09) (Fig. 15).
Recrystallization
122
16
25
40
52
103
198
12
23
39
49
87
0
50
100
150
200
250
300
900 950 1000 1050 1100 1150
Deformation temperature [C]
A
v
e
r
a
g
e
a
r
e
a
o
f
g
r
a
i
n
m
2
]
0.1 s
1.0 s
Strain rate
-1
-1
Soaking parameters: 1100C/2h
Fig. 14. The effect of deformation temperature on the average area of recrystallized grain
after torsion at a rate of 0.1 and 1.0 s
-1
. Initial soaking: 1100C/2h
1,31
1,09
0,99
1,12
1,09
1,04
1,02
1,03
1,00
1,07
1,00
0,6
0,7
0,8
0,9
1
1,1
1,2
1,3
1,4
1,5
1,6
900 950 1000 1050 1100 1150
Deformation temperature [C]
E
l
o
n
g
a
t
i
o
n
c
o
e
f
f
i
c
i
e
n
t
o
f
g
r
a
i
n
s
o
0.1 s
1.0 s
Strain rate
-1
-1
Soaking parameters: 1100C/2h
Fig. 15. The effect of deformation temperature on the elongation coefficient of recrystallized
grain after torsion at a rate of 0.1 and 1.0 s
-1
. Initial soaking: 1100C/2h
19 17
35
48
92
205
12
23
29
41
90
0
50
100
150
200
250
300
900 950 1000 1050 1100 1150
Deformation temperature [C]
A
v
e
r
a
g
e
a
r
e
a
o
f
g
r
a
i
n
m
2
]
0.1 s
1.0 s
Strain rate
-1
-1
Soaking parameters: 1150C/2h
Fig. 16. The effect of deformation temperature on the average area of recrystallized grain
after torsion at a rate of 0.1 and 1.0 s
-1
. Initial soaking: 1150C/2h
The Deformability and Microstructural Aspects of
Recrystallization Process in Hot-Deformed Fe-Ni Superalloy
123
1,08
1,04
1,11
1,04
0,95
1,00
1,09
1,10
1,03
0,94
1,10
0,6
0,7
0,8
0,9
1
1,1
1,2
1,3
1,4
1,5
1,6
900 950 1000 1050 1100 1150
Deformation temperature [C]
E
l
o
n
g
a
t
i
o
n
c
o
e
f
f
i
c
i
e
n
t
o
f
g
r
a
i
n
s
o
0.1 s
1.0 s
Strain rate
-1
-1
Soaking parameters: 1150C/2h
Fig. 17. The effect of deformation temperature on the elongation coefficient of recrystallized
grain after torsion at a rate of 0.1 and 1.0 s
-1
. Initial soaking: 1150C/2h
An increase of the initial soaking temperature to 1150C/2h and alloy deformation within the
range of 9001150C at a rate of 0.1 s
-1
induces a similar reduction in size of the grain plane
section area from 19 m
2
at 900C to 205 m
2
at 1150C (Fig. 16). An increase of the strain rate of
samples to 1.0 s
-1
within the deformation temperature range of 9001100C causes a further
reduction in the grain average area within the range from 12 to 90 m
2
. For both strain rates, the
grains after dynamic recrystallization are approximately equiaxial ( = 0.941.11) (Fig. 17).
A comparison of the size of recrystallized grain in the Fe-Ni alloy after deformation at a
strain rate of 0.1 and 1.0 s
-1
for two variants of initial soaking at 1100C/2 h and 1150C/2 h
is presented in Fig. 18 and 19. After deformation at a strain rate of 0.1 s
-1
within the
temperature range of 9001150C for both variants of initial soaking, a similar grain size
was obtained in the range from 16 to 205 m
2
(Fig. 18). A higher strain rate of 1.0 s
-1
within
the temperature range of 9001100C allows obtaining a slightly higher degree of grain
refining in the range from 12 to 90 m
2
(Fig. 19). Thus, it can be concluded that the initial
microstructure of an alloy after initial soaking has no significant influence on the size of
recrystallized grain after plastic deformation.
16
25
40
52
103
198
19 17
35
48
92
205
0
50
100
150
200
250
300
900 950 1000 1050 1100 1150
Deformation temperature [C]
A
v
e
r
a
g
e
a
r
e
a
o
f
g
r
a
i
n
m
2
]
1100C/2h
1150C/2h
Soaking parameters
Strain rate 0.1 s
-1
Fig. 18. The effect of deformation temperature on the average area of recrystallized grain
after initial soaking of the alloy at 1100C/2 h and 1150C/2 h
Recrystallization
124
12
23
39
49
87
12
23
29
41
90
0
50
100
150
200
250
300
900 950 1000 1050 1100
Deformation temperature [C]
A
v
e
r
a
g
e
a
r
e
a
o
f
g
r
a
i
n
m
2
]
1100C/2h
1150C/2h
Soaking parameters
Strain rate 1.0 s
-1
Fig. 19. The effect of deformation temperature on the average area of recrystallized grain
after initial soaking of the alloy at 1100C/2 h and 1150C/2 h
The average grain plane section area of samples deformed under the same conditions but
with a different initial grain size after initial soaking is similar. The average size of
recrystallized austenite grain depends mainly on the deformation temperature and, to a
lesser degree, on the strain rate applied for the alloy (Fig. 20 and 21).
0
20
40
60
80
100
120
800 850 900 950 1000 1050 1100 1150 1200
Deformation temperature [C]
A
v
e
r
a
g
e
a
r
e
a
o
f
g
r
a
i
n
m
2
]
= 0.0023e
0.0096 * T
R
2
= 0.9812
= 0.0057e
0.0088 * T
R
2
= 0.9601
0.1 s
-1
1.0 s
-1
Strain rate
Soaking parameters: 1100C/2h
Fig. 20. Relationship between the average grain area after recrystallization versus
deformation temperature and strain rate. Initial alloy soaking: 1100C/2 h
The Deformability and Microstructural Aspects of
Recrystallization Process in Hot-Deformed Fe-Ni Superalloy
125
0
20
40
60
80
100
120
800 850 900 950 1000 1050 1100 1150 1200
Deformation temperature [C]
A
v
e
r
a
g
e
a
r
e
a
o
f
g
r
a
i
n
m
2
]
= 0.0028e
0.0093 * T
R
2
= 0.9603
= 0.0088e
0.0083 * T
R
2
= 0.9116
Soaking parameters: 1150C/2h
0.1 s
-1
1.0 s
-1
Strain rate
Fig. 21. Relationship between the average grain area after recrystallization versus
deformation temperature and strain rate. Initial alloy soaking: 1150C/2 h
An assessment of the degree of influence of the deformation temperature and strain rate on the
average grain plane section area of the recrystallized Fe-Ni alloy was obtained after
introducing the Zener-Hollomon Z parameter. As appears from the dependencies developed,
the fine-grained microstructure of the alloy after deformation was obtained more easily after
initial soaking at a temperature of 1100C/2h when compared to soaking at 1150C/2 h (Fig.
22). This is evidenced by lower Z parameter values for the deformation after initial soaking at
1100C/2 h: a low strain rate and a moderate deformation temperature.
0
20
40
60
80
100
120
140
160
180
200
220
34 36 38 40 42 44 46 48 50 52 54
ln Z
A
v
e
r
a
g
e
a
r
e
a
o
f
g
r
a
i
n
m
2
]
Soaking parameters
1150C/2h
Soaking parameters
1100C/2h
R
2
=0.9138
R
2
=0.8833
Fig. 22. Relationship between the Zener-Hollomon parameter and the average grain area
after recrystallization. Initial alloy soaking: 1100C/2 h and 1150C/2 h
Recrystallization
126
For both variants of initial soaking of the alloy, relationships were determined between the
average plane section area of the recrystallized grain and the Z parameter (eq. 16 and 17):
- for the alloy after initial soaking 1100C/2h:
= 9.3 10
5
Z
-0.250
[m
2
] (16)
- for the alloy after initial soaking 1150C/2h:
= 2.0 10
6
Z
-0.224
[m
2
] (17)
The determined relationships are essential where the critical value of a significance test for
the direction factor of regression line "p" is less than 0.05. In the analyzed case, the "p" factor
value for both variants of initial soaking of the alloy, i.e. at 1100C/2h and 1150C/2h,
equaled 3.0810
-5
and 5.0310
-5
, respectively, which indicates the significance of the
determined relationships.
3.4 Substructure of hot-deformed alloy
The recovery and dynamic recrystallization processes during hot plastic deformation of the
Fe-Ni alloy cause changes in the dislocation and subgrain substructures. An analysis of the
microscopic examination results allows affirming that the nature and extent of changes in
the alloy substructure was dependent on the deformation temperature and strain rate, and
the conditions of initial soaking. After initial soaking (1100C/2 h) and deformation at a
temperature of 900C and a rate of 0.1 and 1.0 s
-1
, effects of dynamic recovery and dynamic
recrystallization (Fig. 23 and 24) were observed in the alloy microstructure. In grain areas
with major defects of the austenite, a cellular dislocation substructure and subgrains with
different dislocation densities were formed.
Fig. 23. Alloy microstructure after soaking at 1100C/2 h and deformation at 900C/0.1 s
-1
.
Subgrains and recrystallized grains
The Deformability and Microstructural Aspects of
Recrystallization Process in Hot-Deformed Fe-Ni Superalloy
127
Fig. 24. Alloy microstructure after soaking at 1100C/2 h and deformation at 900C/1.0 s
-1
.
Subgrain structure formation
At a higher deformation temperature, e.g. 950C, the proceeding recovery processes were
accompanied by intensive dynamic recrystallization (Fig. 25 and 26). In the alloy
substructure, dynamically recrystallized grains were observed next to the subgrains
(Fig. 25). In the samples deformed at a higher rate, 1.0 s
-1
, at this temperature the fraction
and size of dynamically recrystallized microregions increase (Fig. 26).
The alloy deformed within the temperature range of 10001050C is characterized by a
microstructure typical of a dynamically recrystallized material (Fig. 27 and 28). The
austenite microstructure is composed predominantly of recrystallized grains free of
dislocations (Fig. 27). Further perfecting of the substructure is observed in the neighbouring
subgrains, as evidenced by equiaxiality of the subgrains and the decreasing density of
dislocations inside them (Fig. 28). It was found that a higher strain rate (1.0 s
-1
) leads to a
growth of the subgrain and a reduction of the dislocation density.
Fig. 25. Alloy microstructure after soaking at 1100C/2 h and deformation at 950C/0.1 s
-1
.
Subgrains and recrystallized grains
Recrystallization
128
Fig. 26. Alloy microstructure after soaking at 1100C/2 h and deformation at 950C/1.0 s
-1
.
Regions of recrystallized austenite and subgrains
Fig. 27. Alloy microstructure after soaking at 1100C/2 h and deformation at 1000C/0.1 s
-1
.
Recrystallized grains with twins and subgrains
Fig. 28. Alloy microstructure after soaking at 1100C/2 h and deformation at 1050C/1.0 s
-1
.
The process of further perfecting of the subgrain structure
The Deformability and Microstructural Aspects of
Recrystallization Process in Hot-Deformed Fe-Ni Superalloy
129
In the alloy substructure, after deformation at the highest temperature of 11001150C, a
reincrease of the dislocation density and repolygonization were observed (Fig. 29). At a
higher strain rate of 1.0 s
-1
, areas with recrystallized grain free of dislocations are dominant
in the alloy substructure (Fig. 30).
An increase of the initial soaking temperature to 1150C/2h inhibits the dynamic recovery
and recrystallization processes (Fig. 31 and 32). In the regions with austenite subgrain after
deformation at a temperature of 900C at a rate of 0.1 s
-1
, deformation microtwins appear
(Fig. 31). The subgrains being formed, especially at a high strain rate of 1.0 s
-1
, have an
elongated shape and different dislocation densities (Fig. 32).
At a higher deformation temperature, e.g. 1000C, and a strain rate of 0.1 and 1.0 s
-1
, the
alloy substructure reconstruction is inhomogeneous. It is characterized by the presence of
areas where structural changes are inhibited and accelerated (Fig. 33 and 34).
Effects of dynamic recovery (Fig. 33) and dynamic recrystallization (Fig. 34) were found
there. The growth of new grains in the dynamic recrystallization process proceeded through
the coalescence of subgrains and their subsequent growth (Fig. 35). The bending of the grain
boundary towards areas with a higher dislocation density indicates the direction of the
boundary movement. The principal mechanism of the coalescence includes reactions
between dislocations which lead to disappearance of the dislocation boundary and
formation of grain from the combination of several neighboring subgrains.
The dislocation density in the subgrain area does not decrease significantly when increasing
the deformation temperature to 1100C compared to a lower deformation temperature
(Fig. 36 and 37). The dynamic deformation phenomena are accompanied by a continuous
process of structural reconstruction of the material, i.e. repolygonization. It consists in
re-saturation of subgrains with dislocations and their rearrangement together with the
creation of new subboundaries and walls of a polygonal type (Fig. 36).
Deformation of the alloy at a higher rate of 1.0 s
-1
was accompanied in the substructure by
dislocation rearrangement with the formation of polygonal walls and a cellular substructure
(Fig. 37). The phenomenon of repolygonization in the austenite grains was observed at the
highest deformation temperature (11001150C).
Fig. 29. Alloy microstructure after soaking at 1100C/2 h and deformation at 1100C/0.1 s
-1
.
The effects of repolygonization in austenite subgrains
Recrystallization
130
Fig. 30. Alloy microstructure after soaking at 1100C/2 h and deformation at 1100C/1.0 s
-1
.
Large austenite subgrains with a low dislocation density
Fig. 31. Alloy microstructure after soaking at 1150C/2 h and deformation at 900C/0.1 s
-1
.
Subgrain structure and grains with microtwins
Fig. 32. Alloy microstructure after soaking at 1150C/2 h and deformation at 900C/1.0 s
-1
.
Non-equiaxed, austenite subgrains
The Deformability and Microstructural Aspects of
Recrystallization Process in Hot-Deformed Fe-Ni Superalloy
131
Fig. 33. Alloy microstructure after soaking at 1150C/2 h and deformation at 1000C/0.1 s
-1
.
The effects of polygonization and deformation microtwins
Fig. 34. Alloy microstructure after soaking at 1150C/2 h and deformation at 1000C/1.0 s
-1
.
The grains after dynamic recrystallization
Fig. 35. Alloy microstructure after soaking at 1150C/2 h and deformation at 1000C/0.1 s
-1
.
The recrystallized grain formation as a result of coalescence of subgrains
Recrystallization
132
Fig. 36. Alloy microstructure after soaking at 1150C/2 h and deformation at 1100C/0.1 s
-1
.
The formed subgrain structure with dislocations
Fig. 37. Alloy microstructure after soaking at 1150C/2 h and deformation at
1100C/1.0 s
-1
. Austenite repolygonization and a cellular dislocation structure
The course of changes in the subgrain sizes depending on the temperature deformation and
strain rate for the two variants of initial soaking of the alloy is shown in Fig. 38 and 39. It
was found that an increase in the alloy deformation temperature from 900 to 1150C results
in a growth of the subgrain.
The Deformability and Microstructural Aspects of
Recrystallization Process in Hot-Deformed Fe-Ni Superalloy
133
0
1
2
3
4
5
6
7
8
9
10
850 900 950 1000 1050 1100 1150 1200
Deformation temperature [C]
A
v
e
r
a
g
e
s
u
b
g
r
a
i
n
s
i
z
e
m
2
]
Strain rate: 1.0 s
-1
Strain rate: 0.1 s
-1
R
2
= 0.8765
R
2
= 0.9280
Soaking parameters: 1100C/2h
Fig. 38. The effect of temperature deformation and strain rate on the average subgrain size.
Initial alloy soaking: 1100C/2 h
0
1
2
3
4
5
6
7
8
9
10
850 900 950 1000 1050 1100 1150 1200
Deformation temperature [C]
A
v
e
r
a
g
e
s
u
b
g
r
a
i
n
s
i
z
e
m
2
]
Strain rate: 0.1 s
-1
Soaking parameters: 1150C/2h
Strain rate: 1.0 s
-1
R
2
= 0.9721
R
2
= 0.9126
Fig. 39. The effect of temperature deformation and strain rate on the average subgrain size.
Initial alloy soaking: 1150C/2 h
However, no influence was observed of the conditions of initial soaking on the subgrain
size. The average area of the subgrain plane section varied within the range from 1.0 m
2
to 7.8 m
2
for both variants of initial soaking. The influence of the strain rate on the subgrain
size was more significant, in particular for the initial soaking at 1100C/2 h (Fig. 38). More
intensive changes in the subgrain size were observed at a low strain rate of 0.1 s
-1
, which can
be explained by a higher cumulative deformation in the samples.
Recrystallization
134
The dislocation density depending on the deformation temperature and strain rate for the
two variants of initial soaking of the alloy is shown in Fig. 40 and 41. An increase of the
deformation temperature was accompanied by a decreasing dislocation density. No
significant influence was found of the initial soaking parameters on the dislocation density.
R
2
= 0.9812
0
0,5
1
1,5
2
2,5
3
3,5
4
4,5
5
850 900 950 1000 1050 1100 1150 1200
Deformation temperature [C]
A
v
e
r
a
g
e
d
i
s
l
o
c
a
t
i
o
n
d
e
n
s
i
t
y
1
0
1
3
[
m
-
2
]
Strain rate: 0.1 s
-1
Strain rate: 1.0 s
-1
R
2
= 0.9141
Soaking parameters: 1100C/2h
Fig. 40. The effect of temperature deformation and strain rate on the average dislocation
density. Initial alloy soaking: 1100C/2 h
0
0,5
1
1,5
2
2,5
3
3,5
4
4,5
5
850 900 950 1000 1050 1100 1150 1200
Deformation temperature [C]
A
v
e
r
a
g
e
d
i
s
l
o
c
a
t
i
o
n
d
e
n
s
i
t
y
1
0
1
3
[
m
-
2
]
Strain rate: 0.1 s
-1
Soaking parameters: 1150C/2h
Strain rate: 1.0 s
-1
R
2
= 0.9614
R
2
= 0.8839
Fig. 41. The effect of temperature deformation and strain rate on the average dislocation
density. Initial alloy soaking: 1150C/2 h
For both variants of initial soaking, the dislocation density varied within a narrow range
from 0.810
13
m
-2
to 2.510
13
m
-2
. The gradual reduction in the dislocation density observed
in the samples as the deformation temperature increased from 900 to 1150C shows a
The Deformability and Microstructural Aspects of
Recrystallization Process in Hot-Deformed Fe-Ni Superalloy
135
continuous process of substructure reconstruction and redeformation. For both variants of
initial soaking of the alloy, higher dislocation densities were obtained for the lower strain
rate (0.1 s
-1
), which can be explained by a higher cumulative deformation in the material.
4. Summary
The work analyzes the relationships between the conditions of hot plastic deformation and
deformability and microstructure of an austenitic Fe-Ni superalloy precipitation-strengthened
by phase ' type. The hot torsion tests carried out in the range of temperature of 9001150C, at
a strain rate of 0.1 s
-1
and 1.0 s
-1
allowed determining the influence of the initial soaking
conditions and deformation parameters on technological plasticity of the Fe-Ni superalloy, as
well as on the strengthening and structure reconstruction processes. It was found that optimal
values of the maximum yield stress
pp
and threshold deformation
f
, as well as the required
fine-grain microstructure, were obtained for the alloy after initial soaking at 1100C/2 h and
deformation at a rate of 0.1 s
-1
in the temperature range of 1050950C. The increase of yield
stress,
pp,
and the decrease of deformability of the alloy,
f
, as the initial soaking temperature
was rising up to 1150C/2 h, with the deformation rate increasing to 1.0 s
-1
, was associated
with a growth of the initial grain size and the degree of austenite saturation with alloying
elements. As a result of these processes, the stacking fault energy (SFE) of the austenite
increased and so did the ability of the material to strengthen. This, in turn, led to an increase of
activation energy of the hot plastic deformation process from the value Q = 441.8 kJ/mol (after
initial soaking at 1100C/2 h) to Q = 518.7 kJ/mol (after initial soaking at 1150C/2 h).
An analysis of the flow curves and the examination results of the Fe-Ni alloy microstructure
and substructure revealed dynamic recovery, recrystallization and repolygonization,
occurring consecutively in the course of hot deformation. None of the detected stages of
changes in the alloy structure constituted an independent process. Their course depended
on both, the deformation parameters (T, c
35
to the rolling plane (Barrett & Jonas, 1997). For instance,
Barnett and Kestens (Barnett et al., 1999) reported that the increasing density and severity of
these in-grain shear bands lead to a bulk recrystallization texture dominated by {1 1 1}<1 1
2> near the normal directionrolling direction (NDRD) for low carbon, ultra low carbon
and interstitial free steels. In the present chromium stabilized EDD steel, the shear bands
were inclined 32
with respect to the rolling plane and were almost parallel to each other
within the individual grains.
Fig. 11. Hot band microstructure of chromium stabilized EDD in hot rolled plates with
flattened ferrite grains closely parallel to the sheet plane and inside the grains, some shear
bands are observed (Mendoza et al., 2004).
Figure 12 shows {1 0 0} pole figures for the steel in the hot rolled, cold rolled and annealed
conditions. It may be observed that in the cold rolled sheet, a {5 5 4}<2 2 5> component was
noticed near {1 1 1}<1 1 2> texture. While, in the annealed sheet an appreciably strong
texture than in the as cold rolled condition was observed from {5 5 4}<2 2 5> to {2 1 1}<0 1
1>. The microstructure obtained in the annealed condition was partially recrystallized,
which was attributed to the presence of chromium-carbides precipitates that helped retard
the recrystallization rate of the low carbon Al-killed/Cr-stabilized steel and which
eventually was instrumental in the formation of {1 1 1}<1 1 2> textures.
Fig. 12. {1 0 0} pole figures for the chromium stabilized EDD steel in (a) hot rolled, (b) cold
rolled and (c) annealed conditions (Mendoza et al., 2004).
Physical Metallurgy and Drawability of Extra Deep Drawing and Interstitial Free Steels
149
The mechanical properties of the annealed chromium stabilized EDD sheet are depicted in
Figure 13. The use of chromium instead of niobium or titanium to stabilize the low carbon
steel was effective in slowing down the recrystallization rate, thus, enhancing the formation
of {1 1 1}<1 1 2> textures and achieving r values (r
m
- values) >2 (Mendoza et al., 2004).
Fig. 13. Mechanical properties of annealed coils. (a) Hardness Rockwell b, (HRb), (b) 0.2% of
yield strength, (YS), (c) tensile strength, (TS), (d) percent of elongation, (El), (e) the average
plastic anisotropy value, (r) and (f) the strengthening hardening exponent, n (Mendoza et
al., 2004).
4.4 Phosphorous effect
It was observed by researchers that rimmed steel with P addition and by annealing the cold
rolled sheet in decarburization furnace improved formability and thus studies were made
on EDD steels to examine the interactive effect of P and AlN precipitation on formability of
the steel sheets (Beranger., et al).
4.4.1 Heating rate effect on phosphorous added EDD steel
Rephosphorized Al-killed (EDD-P) steels show the similar dependence on heating rate of
grains to conventional Al-killed steels (Ono et al., 1982). However, the smaller grain size,
lower grain elongation ratio, lower r
m
-value, faster recrystallizatio rate observed in the EDD-
P steel were all attributed to the effect of phosphorous on the precipitation site of AlN. If
phosphorus weakens the retardation effect of AlN on recrystallization, the restriction on
nucleation of less favourable orientations, other than the {111} nucleus by AlN, which is
usually observed in EDD steels will be relieved, resulting in the reduction of (111} intensity
and an increase in the {100} intensity. In EDD-P steel {111}<112> orientation is observed
unlike {111}<110> in the conventional EDD, which was attributed to the modification of cold
rolled microstructure by phosphorous.
Recrystallization
150
Figure 14 (Ono et al., 1982) shows the change in X-ray integrated intensity during heating.
From the early stage to the half way of recrystallization, a rapid increase in the {222}
intensity in association with a rapid decrease in the {200} and {110} intensities were observed
in the EDD-P steel (Steel1: C: 0.05, Mn-0.24, P:0.069) as well as in the conventional one
(Steel-4: C:0.05, Mn:0.26, P:0.016). However, at the end of recrystallization, the EDD-P steel
showed a lower {222} intensity and a higher {200} intensity than the conventional one. There
was a marginal difference in the {110} intensity between them.
4.4.2 Manganese effect on phosphorous added EDD steel
An optimum combination of phosphorus and manganese content in steel can render a
strong {111} texture through a simulated batch annealing cycle. As per Hu and Goodman
(Hu & Goodman, 1970) and Hughes and Page (Hughes & 1971) the interaction between
manganese and carbon would affect the recrystallization kinetics and thus, recrystallization
textures. Further, Matsudo et al. (Matsudo et al, 1984) reported the detrimental effect of the
Mn-C interaction on the drawability. On the other hand, it is widely known that steels
containing phosphorus are likely to show a banded structure with segregated P-bearing
ferrite, suggesting phosphorus and carbon in steel can act as repulsive elements to each
other.
Fig. 14. Change in integrated intensity of Steels 1 and 4 during continuous heating at a peak
heating rate 50C/hr. (Ono et al., 1982).
Hu (Hu, 1977) reported that the drawability of low-carbon steel containing 0.067 pct P
diminished more gradually than that of P-free steel with increasing manganese content up
to 0.3 pct.
However, the work conducted by Chung et al (Chung et al., 1987) to examine the effect of
manganese on the development of {111} recrystallization textures of P-containing low-
carbon Al-killed steel sheet demonstrated that the phosphorous addition could modify the
Mn-C interaction favourably for achieving improved drawability in EDD steels. Figure 15
shows the (200) pole figures for the 0.1 pct P and P-free steels annealed at 973 K for 3 hours.
The P-containing steels manifested {554} (225)-type texture and the texture was extremely
Physical Metallurgy and Drawability of Extra Deep Drawing and Interstitial Free Steels
151
strong in the 0.78 pct Mn-0.1 pct P steel. On the other hand, in the P-free steels the texture
was not as strong at 0.5 pct Mn as it was at 0.06 pct Mn.
Fig. 15. (200) pole figures for (a) 0.1 pct P steels normalized, cold rolled, and annealed at 973
K for 3 h and (b) P-free steels normalized, cold rolled, and annealed at 973 K for 3 h (Chung
et al., 1987).
Recrystallization
152
To confirm the effectiveness of phosphorus addition to a high manganese steel in order to
get a high r
m
-value, the 0.2 pct P-1.2 pct Mn steel was employed. The recrystallization
texture, as shown in Figure 16, had strong {554}(225) components which were favorable for
a high r
m
-value. The r
m
-value and other mechanical properties are shown in Table 1 (Chung
et al., 1987). The high manganese P-containing steel had a tensile strength of 430 MPa and
an r
m
-value of 2.0.
Fig. 16. (200) pole figure of 0.2 pct P-1.2 pct Mn steel preaged at 843 K for 24 h, cold rolled 75
pct, and annealed at 973 K for 4 h (Chung et al., 1987).
The micrographs of the P-free steel and the P-containing steel hot rolled and aged are shown
in Figures 17 and 18, respectively. In the P-free steels, the precipitates that were identified as
AIN dispersed within the grain at 0.06 pct Mn, whiles the precipitates were barely noticed at
0.5 and 1.0 pct Mn. On the other hand, in the P-containing steels, precipitates which might
be (FeMn)
3
C and/or A1N were dispersed uniformly in the ferrite matrix only at the 1.5 pct
Mn level. From the results, it must be noted that the fine precipitates of nitrogen and/or
carbon disperse uniformly with decreasing manganese in the P-free steels, and vice versa
with increasing manganese in the P-containing steels.
Fig. 17. Transmission electron micrographs of P-free Al-killed steels aged at 833 K for 1000 h
after hot rolling. (a) 0.06 pct Mn, (b) 0.5 pct Mn, and (c) 1.0 pct Mn (Chung et al., 1987).
Physical Metallurgy and Drawability of Extra Deep Drawing and Interstitial Free Steels
153
Fig. 18. Transmission electron micrographs of 0.2 pct P steels aged at 843 K for 24 h after hot
rolling. (a) 0.1 pct Mn, (b) 0.5 pct Mn, and (c) 1.5 pct Mn (Chung et al., 1987).
Table 1. Mechanical properties of 0.2 pct P-1.2 pct Mn Al-Killed steel preaged at 8543 K, cold
rolled 75 pct, and annealed at 973 K for 4 hrs.
Thus, a strong recrystallization texture with {554} (225) texture component and an uniform
dispersion of (FeMn)
3
C in ferrite matrix can be developed through batch annealing of a P-
containing EDD- steel if the steels contain high P and high-Mn .
5. On metallurgy of ultra-low carbon interstitial-free (IF) steels
Interstitial free steels are highly formable due to their low carbon and nitrogen content of
((wt%) < 0.003% and < 0.004%, respectively). The C and N are tied to Ti, Nb etc as
precipitates. A limited excess of titanium or niobium relative to carbon, nitrogen and sulfur
contents has a favourable influence on mechanical properties. During hot strip rolling, the
level of interstitial elements, such as C and N, remnant from the steel making process, can be
reduced by combining them with the stabilizing elements. The application of these steels is
in the rear floor pan, front, rear door inners, etc.
5.1 Titanium stabilized IF steels
Titanium is very effective in scavenging nitrogen, sulphur, and carbon, readily forming TiN
during casting and TiS during slab reheating . Subsequently, while the nitrogen and sulphur
are scavenged, remaining Ti ties up with TiC during coiling. The minimum amount of
titanium required for full stabilization, based on a stoichiometric approach, is (Tither &
Stuart, 1995):
Ti
stab
= 4C+3.42N+1.5S
Recrystallization
154
It was proposed that excess Ti addition than that required to combine with all C, N, and S
was beneficial to achieve high r values (Gupta & Bhattacharya, 1990). Excess Ti (Ti*) is
given by :
Ti*=Ti
total
-(4C+3.42N+1.5S)
However, it is also commented that an excess Ti content can be linked to the incidence of
surface streaking. The frequency of appearance of this type of defect can be minimized using
Nb in combination with Ti. Titanium only intierstitial free steels are the least susceptible to
compositional changes and process variation. This is attributed to the precipitation of Ti
compounds at higher temperatures (TiN, TiS) that play less of a role in subsequent lower
temperature processing (Krauss et al., 1991). Tsunoyama et al (Tsunoyama et al, 1988)
considered three precipitation mechanisms suggested by various authors: a) TiS may
provide a preferential site for TiC nucleation. Thus, lowering the S content can retard the
precipitation of TiC b) with decreasing Ti content, precipitation of Ti
4
C
2
S
2
occurs in place of
TiS and controls carbon stabilization. The stabilization of C by Ti
4
C
2
S
2
is preferred compared
to TiC precipitation as Ti
4
C
2
S
2
precipitates are larger, remove solute carbon from the matrix
earlier, and are more stable.
The precipitation mechanism of titanium in these steels is summarized in Figure 19 (Tither,
1994). It is postulated that titanium nitride particles formed during slab casting acted as
nucleation sites for the precipitation of TiS and Ti
4
C
2
S
2
. The small amount of carbon
remaining is precipitated as TiC. During reheating of the slab, solution of carbosulphide
occurs, leaving only TiS and TiN. Cooling of the strip to the inctercritical (austenite -ferrite
transformation) temperature region during hot rolling transforms TiS to Ti
4
C
2
S
2
by the
absorption of titanium and carbon.
5.2 Niobium stabilized IF steels
In Nb IF steels, Nb forms carbonitride precipitates. The aluminium addition during the
killing process also reacts with N to form AlN. However, the favourable solubility product
of AlN compared to NbCN, leads to preferential precipitation of AlN at a higher critical
temperature. This reduces the N available for the precipitation as NbCN. The solubility
products are dependent on bulk chemistry, temperature, and precipitate composition. The
precipitates sequence in such steels can be considered as (Holie, 2000)
Al+N=AlN
Mn+S=MnS
Nb+C+N=NbC
The solute Nb on grain boundaries introduces site competition for the elements like
phosphorous, which enables lower ductile to brittle transition temperatures, poorer
elongation and r
m
- values than Ti IF steels. The niobium steels have higher recrystallisation
temperature (750 - 800C) than titanium stabilized IF steels (Baker et al., 2002) and it was
reported that the temperature of recrystallisation of Nb IF could be up to 50 K higher than
that of the Ti IF steels. This in turn requires higher finish rolling and annealing
temperatures. The higher recrystallization temperature is attributed to the low temperature
Physical Metallurgy and Drawability of Extra Deep Drawing and Interstitial Free Steels
155
formation of niobium carbides, producing fine particles that readily retard grain boundary
movement during annealing.(Tsunoyama, 1990). The temperature can be reduced by
lowering C content. In addition, recrystallisation temperature can be reduced to enhance
drawability, using higher coiling temperature, so that precipitates are coarsened and
subsequent pinning effect is reduced.
Fig. 19. Schematic mechanism of carbosulphide and carbide formation in titanium stabilized
IF steels (Tither et al., 1994).
5.3 Titanium and niobium stabilized IF steels
Titanium in association with niobium added IF steels have the advantages of both Ti only
and Nb only IF steels with reduced disadvantages. This combination provides the best
combination of properties (Tokunaga 1990). The levels of Ti and Nb additions that are
required to stabilize Ti +Nb IF steels are given by (Holie, 2000; Tokunaga & Kato, 1990)
(%Nb)-(7.75(%Ti)-3.42(%N)-1.5(%S))/4=0
If insufficient Ti is added such that the level of addition falls below that required for
stabilization, TiN forms first and any S that is left behind after the formation of TiS is
consumed as MnS. If the steel is understabilised by the insufficient addition of Ti and/or Nb
then excess C exists in solid solution and a bake hardenable steel may be produced.
For IF steels, the increase in r
m
-value with degree of cold work is continuous over the whole
range employed in industrial production (Beranger et al).An increase in annealing
temperature and/or holding time improves the deep drawability due to grain growth and
recrystallization texture enhancement. Further, these grades have very high Ac1
temperatures, enabling continuous annealing at temperatures of ~850C, or even higher.
For such steels, the metal has no yield point elongation and no requirement of overaging
treatments in continuous annealing, due to the absence of interstitials C and N.
Recrystallization
156
5.4 Phosphorous effect on IF steels
One of the ways of increasing the strength of the IF steels is addition of solid solution
strengtheners viz. P, Si and Mn to the steel (Rege et al., 2000). It has been found that P is the
most potent and cost effective solid solution strengthener that increases strength without
appreciably affecting the drawability of the steel (Katoh et al., 1985; Tokunaga & Kato,
1990). Mn, on the other hand, significantly deteriorates drawability and ductility (Irie et al.
1981; Katoh et al., 1985; Tokunaga & Kato, 1990), and Si adversely affects coating adhesion
(Nishimoto et al., 1982). Thus, P is the preferred addition to increase the strength of IF-steel.
However, it is found that phosphorus tends to segregate at the grain boundaries or
precipitate out of the matrix during the recrystallization annealing. The phosphorus
segregations reduce the cohesive strength of the grain boundaries, which leads to the
secondary cold work embrittlement (CWE) and reduces the resistance of the steel to brittle
fracture or more precisely, intergranular fracture (El-Kashie et al. 2003; Rege & DeArdo,
1997). The CWE is defined as the susceptibility of the sheet material to intergranular fracture
during the secondary work of deeply drawn part or while in service. Cao et al (Cao, 2005)
had reported that the segregation of phosphorus occurred when the P>0.07 wt pct in the IF
steel. It is also found that the batch annealing leads to the higher phosphorus than the
continuous annealing. The segregation behavior of phosphorus in the Ti and Ti+Nb IF steels
was studied by Rege et al (Rege et al., 2000). It was found that the segregation of
phosphorus to the ferrite grain boundaries occurred not only during the coiling stage of the
thermo-mechanical processing, but also during the cold rolling and annealing process. The
steels with higher phosphorus content showed higher ductile to brittle transition
temperature, i.e, lower resistance to cold work embrittlement (CWE). The problem of CWE
is more often encountered while annealing time is long, which enables P to segregate in the
grain boundary (Yasuhara, 1996).
The CWE can be avoided by :
controlling the chemical compositioni. e. by partial stabilization of carbon in IF steels
or by addition of B/Nb - It is believed that P and C both segregate to the grain
boundary and compete for the available sites, and C reduces the grain boundary
segregation and embrittlement by P. Furthermore, C enhances the grain boundary
cohesion and counteracts the embrittlement this way. B also plays the same role as that
of C. Hence P-C or P-B site competitive process was expected to minimize the CWE
phenomena. In Nb added steel it is believed that CWE is lower than that in Ti-stabilized
steel that has been attributed to the partial stabilization of C in Nb-containing steel.
grain boundary engineeringi. e. by grain boundary character distributionby
suitable annealing cycles low angle and low -CSL boundaries can be produced by
avoiding the development of continuous random boundary network and which help
reduce CWE.
6. Recrystallization texture and drawability of IF steels
Cold rolling plays an important role in the formation of favourable textures for deep
drawing during annealing, however, has little effect on other properties. The variation of
r-value with cold reduction in the three common IF steel types, Ti-stabilized, Nb-
stabilized and Ti-Nb stabilized is shown in Figure 20 (Tokunaga & Yamada, 1985). Ti -
Physical Metallurgy and Drawability of Extra Deep Drawing and Interstitial Free Steels
157
Nb steels show the highest r-value for equivalent cold reduction. This was attributed to
the fact that the precipitates formed during hot rolling and coiling were not large enough
to compromise the r-value during annealing. A reduction of 90% produced the highest r-
value in all the steels, however, these reductions were rarely achieved in practice, 80%
being more common.
Fig. 20. Effect of cold reduction on r-value for titanium, titanium-niobium, and niobium IF
steels (Tokunaga & Yamada, 1985).
Mendoza et al (Mendoza, 2000) studied the hot rolled precipitation behaviour of Ti-
stabilized IF steels and mechanical properties and reported about improved r
m
-value of 2-
2.14. While Banerjee et al (Banerjee et al., 2008) vividly studied the precipitation of hot-
band and annealed Ti stabilized IF steels to optimize hot and cold spot temperatures during
annealing for improving drawability of the steels to >1.9. The hot bands of the as-received
steels showed two types of texture development: (a) shear texture, manifested by
{225}<554>, {110}<100>, and {113}<332> and (b) cold deformation type texture with a strong
and incomplete - fiber consisting of the components {001}<110>, {112}<110>, and
{111}<110>, and the -fiber components, {111}<110> {111}<112>. The presence of a
substantial amount of coarse equiaxed ferrite grains at the surface and finer grains at the
central region was attributed to the outcome of finish rolling in the two-phase + region.
On the other hand, the deformed grains at the surface and finer grains at the central region
another hot band was the result of rolling in the single- phase region. Figure 21 (Banerjee
21 et al., 2008) shows the 2= 45
o
orientation distribution functions for various combinations
of cold reduction and batch annealing temperatures to obtain improved r
m
-value in an
optimized processing condition. The poor texture and r
m
-values at the annealing
temperature of 660
o
C were associated with the precipitation of fine precipitation of FeTiP-
type within the grains and at the grain boundaries. This study had shown that for the given
chemical composition of the Ti-IF steel, the optimized condition for cold spot temperature in
batch-annealing cycle was 680
o
C, preceded by 80 pct cold reduction, which resulted in an
r
m
-value of as high as 2.29.
Recrystallization
158
The influence of the texture development in Ti-added (0.03, 0.05 and 0.07 wt%) IF on r
m
-
value was investigated by Kim et al (Kim et al., 2005). It was intended to determine the
optimized Ti content for the promotion of deep drawability in the IF steels. For the IF steel
with the composition of 0.0025C, 0.070Mn, 0.002N and 0.007S, the optimum Ti content was
found to be 0.05wt%.
Juntunen et al (Juntunen et al., 2001) investigated the continuous annealing parameters in
laboratory scale on drawability of Ti+Nb stabilized IF and IF-HS steels and it was reported
that r
m
-value could be enhanced by about 13% simply by adjusting the annealing conditions.
The annealing cycle with maximum studied temperature produced the sharpest -fiber and
highest r
m
-value. While, Ruiz-Aparicio et al (Ruiz-Aparicio, et al., 2001) studied the
evolution of the transformation texture in two Ti
4
C
2
S
2
-stabilized interstitial-free (IF) steels
(Ti and Ti/Nb) as a function of different thermomechanical processing parameters. Analysis
showed that the Ti
4
C
2
S
2
-stabilized steels stabilized were not very sensitive to the processing
conditions employed in the study. The study also revealed that, under conditions of large
deformations and coarse austenite grain sizes, the main components of the transformation
textures are the beneficial {111}||ND orientations.
Fig. 21. ODFs of 2 = 45
o
section at various processing conditions for a particular batch
annealing temperature at various cold reductions. Maximum intensity for 660
o
C: 75 pct-
3.53, 80 pct -8.43, and 85 pct-6.43; for 680
o
C (hot spot): 70 pct-9.68, 80 pct-12.13, and 90 pct-
10.38; for 710
o
C: 75 pct-2.95, 80 pct-8.09, 85 pct-6.89, and 90 pct-4.13; and for 750
o
C: 75 pct-
2.0, 80 pct-10.77, 85 pct-9.46, 90 pct-3.32 (Banerjee et al., 2008).
The effects of electric field annealing on the development of recrystallization texture and
microstructure in a Ti+Nb stabilized cold rolled IF steel sheet were studied by He et al (He
et al., 2003) means of ODF analysis and optical microscopy to assess the drawability
response of the steel. Specimens of size 50 mm X 20 mm were cut from the sheet with the
Physical Metallurgy and Drawability of Extra Deep Drawing and Interstitial Free Steels
159
longitudinal direction parallel to the rolling direction. They were then subjected to
isothermal annealing at different temperatures ranging from 650 to 800
o
C for 15 min,
respectively with or without a DC electric field of 200 V/mm. The annealing treatments
were done in a nitrogen atmosphere and at a heating rate of 5
o
C/min to the chosen peak
temperatures. The external electric field was applied by placing the specimens (positive
electrode) in the middle of two parallel stainless steel sheets (negative electrode) that were 2
cm apart. The experimental arrangement is shown in Figure 22.
Fig. 22. Experimental arrangement for annealing with electric field (He et al., 2003).
Heating rate annealing experiments were carried out by Muljono et al (Muljono et al., 2001)
to study the effect of heating rate on the recrystallization kinetics, grain size and texture of
steels with a range of carbon levels (0.0030.05% C). The steels were cold-rolled to 70%
reduction and subsequently annealed at heating rates from 50 to 1000C/sec to
temperatures in the range 600 to 900C. Figure 23 shows that, for the 0.02 and 0.003% C
steels, the {111}||ND texture increases in strength with increased heating rates up to
200C/sec and maintains a plateau thereafter. Both steels exhibit similar trends and only the
relative strength of the -fibre differs. The grain size and texture results are in general
agreement with work by Hutchinson and Ushioda (Hutchinson & Ushioda, 1984). The
{111}||ND components of the recrystallization texture increased at rates up to 200C/sec
due to enhanced nucleation at grain boundary sites.
Fig. 23. Effect of heating rate on strength of recrystallization texture, given as the area
fraction of grains within 15 of{111}||ND (Muljono et al., 2001).
Recrystallization
160
During electric field annealing, specimens work as anode and the applied electric field
reduces the lattice defect energy by lowering the shield effect that would decrease the
driving force for recrystallization (Cao et al.,1990; Conrad, 1989; Wang, 2000). Thus,
although the application of electric field generally reduces the driving force for nucleation
and grain growth, the nuclei with random orientations are much more restricted than that
of the -nuclei. Consequently, the electric field annealing yields a high nucleation rate for the
-nuclei that lead to a relatively strong -texture after complete recrystallization. From
Figures 24 and 26 (He et al, 2003) it can be noted that both kinds of specimens annealed with
and without application of electric field, exhibited a similar tendency in the development of
recrystallization textures, i.e. the -fiber was weakened and the -fiber was strengthened
with increasing annealing temperature. In addition the Figures 4 & 6 also depict that the
application of electric field (200 V/mm) during annealing may promote the development of
the -fiber (ND||<111>) recrystallization texture of the cold-rolled IF steel sheet, which is
beneficial to the deep-drawability. While, Figures 25 and 27 (He et al, 2003) indicate that the
recrystallization was noticeably retarded intensively by electric field annealing under the
investigated conditions.
Fig. 24. = 45
o
sections of the ODFs (levels: 1; 2; 3; . . .) for the specimens annealed without
an electric field at (a) 650
o
C, (b) 700
o
C, (c) 750
o
C and (d) 800
o
C (He et al, 2003).
In an innovative finding, Jeong communicated (Jeong, 2000) that r value was markedly
improved by reducing the carbon content from 0.0035 (Steel B) to 0.0009 pct. (Steel A) in Ti
stabilized IF steels that were treated with 0.25 pct Si, 1.25 pct Mn, and 0.09 pct P to attain
tensile strength of 400 MPa (Figure 28a). The difference in r
m
-value between two steels is 0.2
to 0.3 at all annealing temperatures (Figure 28b). Steel A containing 0.0009 pct carbon
showed a high r
m
-value of about 1.6 while annealed at 800
o
C to 860
o
C, corresponding to
deep drawing quality (DDQ). The r
m
-value increased to 1.85 for extra deep drawing quality
(EDDQ) grade with increasing annealing temperature to 890
o
C. It was thus a remarkable
finding as the steel was high strength steel with tensile strengths of 400 MPa or higher. This
result indicated that while the carbon content decreases below 0.001 pct, superior
formability of EDDQ grade could be achieved in high strength steel with tensile strength of
400 MPa or higher.
The highest r (r
m
-values) in steels A and B were obtained in the specimens annealed at
890
o
C while the coarsening of the ferrite grain was remarkable (Figure 28c,d). In order to
find out the reason for the effect of carbon on the r
m
-value, (200) pole figures for the
annealed sheets were measured. The comparison of (200) pole figures of steels A and B
showed that with the decrease in the carbon content, {554}(225} near the ND//(1l1} texture
became stronger that was responsible for the improvement of r
m
- value (Jeong, 2000).
Physical Metallurgy and Drawability of Extra Deep Drawing and Interstitial Free Steels
161
Song et al (Song et al., 2010) worked on phosphorous segregation and phosphide
precipitation on grain boundaries in association with drawability of rephosphorised IF steel
.The cold rolled steel was annealed at 810
o
C for 90 to 600 sec in a protected environment.
The Table 2 (Song et al., 2010) illustrates yield strength, tensile strength and r
m
-value reduce
while n (work hardening index) value increases with the annealing time from 180 to 600 s.
The phosphorus concentration at grain boundary is shown in Figure 29 (Song et al., 2010). It
was observed that as annealing time increased, the phosphorus concentration increased
from 0.22 to 0.8 wt pct. The phosphorus concentration at the grain boundary is 20 times
higher than that in the matrix for the sample annealed for 600 s, which greatly reduces the
steel strength.
Fig. 25. Microstructures of the specimens annealed without an electric field at (a) 650
o
C, (b)
700
o
C, (c) 750
o
C and (d) 800
o
C (He et al, 2003)
Fig. 26. = 45
o
sections of the ODFs (levels: 1; 2; 3; . . .) for the specimens annealed with an
electric field at (a) 650
o
C, (b) 700
o
C, (c) 750
o
C and (d) 800
o
C (He et al, 2003).
Fig. 27. Microstructures of the specimens annealed with an electric field at (a) 650
o
C, (b) 700
o
C, (c) 750
o
C and (d) 800
o
C (He et al, 2003).
Recrystallization
162
Steel A
Steel B
a)
b)
c)
d)
Fig. 28. Effects of carbon and annealing temperature on (a) tensile strength (b) r (r
m
-value)
and (c) & (d) ferrite grain sizes after annealing at 890
o
C for Steels A and B (Jeong, 2000).
Table 2. Mechanical properties of the annealed samples for different times
Fig. 29. Phosphorus content at grain boundaries in the rephosphorised IF steels annealed for
different time 5 (Song, et al., 2010)
Physical Metallurgy and Drawability of Extra Deep Drawing and Interstitial Free Steels
163
Kang et al (Kang et al., 2007) studied on the effects of aluminium on interstitial free high
strength steel for the improvement of drawability (Figure 30). 78%cold rolled samples
were annealed in an infrared-ray heating furnace. The annealing cycle consisted of
heating the specimens to 830C at a constant heating rate of 7C/s and held at
temperature for 30 s and then cooled to RT. Aluminum content more than 0.10 wt%
improved the formability of the IF-HS. Texture analyses showed that the {111}||ND fiber
(-fiber) was intensified, and <110>||RD (-fiber) was weakened, with the increase of
aluminum content. Recrystallization was completed earlier in the steel with the high
aluminum content and the grain size of the annealed sample was larger than the steel
containing lower aluminium. It was confirmed thorough the SANS analysis that the size
of the precipitates in the sample with higher aluminum content was larger and their
number was much fewer than in the sample with lower aluminum content. It appears that
the high aluminum content in IF-HS containing Mn, P, Ti and Nb improved the
scavenging effect of Ti or Nb and thus purified the iron matrix.
Fig. 30. Effect of aluminum content on mean r-value of the interstitial free high strength steel
containing Mn, P, Ti and Nb (Kang et al., 2007) (Kang et al., 2007)
7. Metallurgy of bake hardening steels
Bake hardening is a diffusion controlled process involving the migration of solute carbon
atoms within the iron lattice. The diffusion of these atoms is affected by heat treatment time
and temperature and the amount of solute present in the steel. Factors such as grain size and
dislocation density may also have an influence.
7.1 Mechanism of bake hardening
The yield strength increase from bake hardening is accompanied by the return of the yield
point and yield point elongation; there may also be a slight increase in tensile strength and
decrease in elongation. To attain such increased strength, the following criteria must be met:
(Matlock et al., 1998)
Recrystallization
164
i. mobile dislocations must be present in the steel.
ii. there must be sufficient concentration of solute in the steel to pin these dislocations.
iii. the solute must be mobile at the aging temperature.
iv. dislocation recovery must be sufficiently slow to prevent softening.
The driving force for pinning is a reduction in lattice energy. Both impurity atoms and
dislocations induce lattice strains in the iron matrix and these strains can be relaxed if the
interstitial atoms diffuse to the vicinity of dislocations (Mizui et al., 1990).Bake hardening
steel is amply used by the automotive industry for the outer panel of cars. This steel grade is
characterized by low yield strength prior to final manufacturing process and a remarkably
enhanced yield strength of the finished product in association with excellent deep
drawability. The increase in yield strength during bake hardening of steel occurs due to the
blocking of otherwise mobile dislocations by forming Cottrell atmosphere of solute carbon
or iron-carbide precipitates during the baking operation after painting at 170
o
C for about 20
minutes (Kozeschnik et al., 1999).
There are several types of bake hardening steel in production today. Individual grades are
determined by the processing technology available and the properties required in the final
product. These steels can be loosely grouped into two main categories, bake hardening EDD
(aluminium killed) steels and bake hardening IF steels.
7.2 Bake hardening EDD steels
The EDD steel grades contain carbon content of the level~ 0.01%. If the carbon is allowed to
remain in solution, room temperature aging occurs. Thus, carbon levels must be controlled
by suitable annealing practices. These steels can exhibit problems during galvannealing,
however, their main disadvantage is poor formability due to the presence of relatively high
carbon contents (Baker et al., 2002). Careful chemistry control during steelmaking is
therefore crucial to ensure suitable amounts of carbon remain in solution in the final
product.
During slow cooling of batch annealed EDD grade steels almost complete precipitation of
the carbon occurs and thus the remaining solute carbon is insufficient to cause bake
hardening. To obtain bake hardening effect in batch annealed EDD, very low carbon grades
of C~ 0.01% with some elements, like phosphorous, that increase carbon concentration in
solution are employed. (Beranger et al.).
While in the case of continuous annealing, the carbon level in solid solution at the end of
annealing is quite high and the solute carbon can produce strengthening (70-80 MPa) by
bake hardening (steel book). Care must be taken, however, with overaging practice during
annealing, to ensure an appropriate amount of carbon (15-25 ppm) is left in solution in the
final product.
7.3 Bake hardening IF steels
IF steels contain very low amounts of total carbon (~0.004%) (Baker et al., 2002).. In these
steels all the interstitial elements are removed from solution by addition of carbide and
nitride formers such as aluminium, titanium, niobium. These steels do not exhibit bake
hardening as they have no interstitial elements in solution. However, the chemistry and
Physical Metallurgy and Drawability of Extra Deep Drawing and Interstitial Free Steels
165
processing of these steels, can be adjusted to leave 15 -25 ppm carbon in solution, to render
bake hardening effect for increasing strength by 30 - 60 MPa . There are several types of IF
steels and in the following sections Ti-only, Nb-only and Ti+Nb steels will be discussed for
bake hardening.
7.3.1 Titanium stabilized interstitial free bake hardening steels
This mechanism of precipitation applies only when some titanium remains in solution: in
just stabilized or understabilized chemistries, formation of TiC and Ti
4
C
2
S
2
is inhibited
because of the low titanium content (Baker et al., 2002). This successive precipitation of
carbides and carbosulphides makes it difficult to control the amount of titanium available
for carbon stabilization and thus the solute carbon content. However, a bake hardening
product can be produced from a titanium chemistry by inhibiting or avoiding TiC
formation, so that the total carbon content remains in solution and is available for bake
hardening.
Work by Tanioku et al. (Tanioku et al., 1991) and Kojima et al. (Kojima et al., 1993) showed
that this can be achieved by controlling total carbon at 15 -25 ppm and titanium at ~ 0.01%.
The manganese content must be kept low (~ 0.3%) to prevent the formation of MnS in
preference to TiS and the slab reheat temperature must be high (~ 1200C) to prevent the
precipitation of Ti
4
C
2
S
2
. Another method for producing titanium IF bake hardening steels,
discussed by Tsunoyama et al. (Tsunoyama et al, 1998) relied on reducing the sulphur level
to minimize formation of TiS, as the role of TiS as a heterogeneous nucleation site for the
precipitation of TiC can lead to the reduction in solute carbon and hence bake hardening
response. Kawasaki et al. (Kawasaki et al., 1991) used a philosophy, involving a reduced
sulphur level (0.005%) in association with an increase in manganese to 1.0% and thus, the
formation of Ti
4
C
2
S
2
was suppressed, leaving carbon in solution and producing a bake
hardening steel.
7.3.2 Niobium stabilized interstitial free bake hardening steels
Niobium is a strong carbide former that can stabilize carbon as NbC when added according
to the stoichiometric ratio:(%Nb)=7.75(%C) (Baker et al., 2002).).
Nitrogen is stabilized by the addition of aluminium to form AlN. This is more stable and
thus forms at higher temperatures than Nb(C,N), so it can be assumed that all niobium is
available for carbide formation. The relative simplicity of carbon stabilization in these steels
makes them ideal for the production of bake hardening grades. Control of solute carbon in
the niobium bearing bake hardening steels can be achieved in two ways. First, insufficient
niobium can be added fully to stabilize the carbon, leaving 15 - 25 ppm in solution after
steelmaking. This methodology requires tight chemistry control during steelmaking and,
because of the presence of solute carbon throughout subsequent processing, the rm-value
can suffer.
The second method requires the full stabilization of carbon during steelmaking. Solute
carbon is then liberated by solution of NbC during annealing. By annealing at high
temperatures (800-850C) and cooling at 420 Ks
- 1
,15 -25 ppm carbon can be retained in
solution (Irie et al., 1982). Since the carbon is fully stabilized until the end of annealing, r
m
-
Recrystallization
166
values in steels of this type are comparable with those of traditional IF grades. High
temperature annealing of this kind can, however, lead to shape problems in the strip such as
heat buckling. Some continuous annealing lines can operate at these high temperatures, but
they are beyond the limits of conventional hot dip galvanizing lines. Steelmakers must
therefore assess their own production capabilities before deciding on a suitable processing
route for the niobium IF based bake hardening steels.
By reducing the sulphur level and increasing manganese, the formation of TiS can be
suppressed, leaving all titanium available for nitrogen stabilization. The precipitation of
Ti
4
C
2
S
2
can also be suppressed in this way, so carbon is controlled by niobium alone. As
with the Nb only compositions, carbon content can be controlled either by
understoichiometric addition of niobium, or by high temperature annealing and controlled
cooling, depending on the capabilities of individual steelmakers. These steels have been
widely researched and are the choice of many manufacturers.
7.3.3 Titanium and niobium stabilized interstitial free bake hardening steels
By reducing the sulphur level and increasing manganese, the formation of TiS can be
suppressed, leaving all titanium available for nitrogen stabilization (Baker et al., 2002).). The
precipitation of Ti
4
C
2
S
2
can also be suppressed in this way, so carbon is controlled by
niobium alone. As with the Nb only compositions, carbon content can be controlled either
by understoichiometric addition of niobium, or by high temperature annealing and
controlled cooling, depending on the capabilities of individual steelmakers. These steels
have been widely researched and are the choice of many manufacturers.
8. Recrystallization texture and drawability of bake hardening steels
Kitamura et al. (Kitamura et al., 1994) had presented a completely different methodology
for the production of titanium based interstitial free bake hardening steels. The theory
suggests that as the steel sheet absorbs carbon by annealing in a carburizing atmosphere, the
Ti/C ratio decreases, eventually resulting in some solute carbon in the matrix. A bake
hardening response of 20 -50 MPa was achieved in this way without compromising r value
(Figure 31) (Kitamura et al., 1994). Through this process, the reduction in r-value due to
solute carbon in solution for interstitial free bake hardening steels can be eliminated as the
steel remains fully stabilized by titanium and the excess carbon is introduced by carburizing
atmosphere during annealing.
Xiaojun and Xianjin (Xiaojun & Xianjin, 1995) developed a new technology (the details
were not mentioned) to improve the drawability of Ti+ Nb stabilized interstitial free high
strength bake hardened steel. The r
m
-value of the experimental sheet treated by the new
technology is as high as 2.67, and this is the highest r
m
-value published so far for
phosphorus-added high strength and deep drawing sheet steels with increased strength due
to bake hardening. Compared to conventional technology (Figure 32a), the new technology
annealing rolling texture (Figure 32b) exhibited strong {111} components and weak {100}.
The crystal orientations corresponding to the peak values of orientation concentrations of
the texture were found to be changed from conventional (111)(112) orientations to (111)(011)
orientations for the new technology (Figure 33).
Physical Metallurgy and Drawability of Extra Deep Drawing and Interstitial Free Steels
167
Fig. 31. Effect of processing on relationship between bake hardenability and r value for Ti-
stabilized IF steel (Kitamura et al., 1994).
Fig. 32. =45
o
sections of ODFs of annealing textures obtained by two technologies
respectively: (a) conventional processing and (b) new processing (Xiaojun & Xianjin, 1995).
Fig. 33. Comparison of annealing textures obtained by two technologies respectively. (a) -
fiber axis textures (b) -fiber axis textures (Xiaojun & Xianjin, 1995).
Recrystallization
168
Bake hardenability and drawability of IF steels with under- to over-stoichiometric atomic
ratios of Ti/N and Nb/C were studied by Storojeva et al
(http://www.cbmm.com.br/portug/sources/techlib/report/novos/pdfs/stabiliza.pdf).
They observed that the r
m
-value increased with a larger ferrite grain
(http://www.cbmm.com.br/portug/sources/techlib/report/novos/pdfs/stabiliza.pdf) thus a
high annealing temperature is favorable for good cold formability. The authors also
reported that the r
m
-value of the steels were higher that had lower solute carbon in the hot
bands (Storojeva et al., 2000). This is confirmed (Figure 34) by just comparing the Ti+Nb
containing steels, where the r
m
-values of the steels with the high solute carbon in hot strip
(12-14 ppm) were remarkably lower than those of the steels without any solute carbon in hot
strip. However, the r
m
-value of Ti-free, just Nb-containing steel with the high solute carbon
content (14 ppm) in the hot strip was almost as high as in the Ti+Nb steels without any
solute carbon. Thus, titanium free IF steel exhibits a lower recrystallization start temperature
and by this means enhances r
m
-value in the final product, allowing compensation of the
negative effect of solute carbon in the hot band. Figure 35 (http://www.cbmm.com.br/
portug/sources/techlib/report/novos/pdfs/stabiliza.pdf) summarizes test results of the r
m
-
value and BH-effect for annealing temperatures up to 840C. It indicates, that a BH-effect
>30 MPa together with r >1.7 can be obtained with the Ti-free, just Nb containing steel.
Fig. 34. r
m
- value of steels with various solute carbon in hot band of IF steels.
Fig. 35. r
m
-value BH-effect for Ti containing IF steels.
Physical Metallurgy and Drawability of Extra Deep Drawing and Interstitial Free Steels
169
9. Recrystallization texture and drawability of warm rolled steels
The demand for thin and ultra thin rolled products led to an intensive search for
alternatives to conventional hot and cold rolling processes. A number of studies were
undertaken to investigate the advantages of rolling at temperatures between 850 and
500C that is known as warm rolling (Harlet et al., 1993). Because of the lower reheat
temperatures, warm rolling leads to lower production costs than hot rolling and requires
significantly lower loads than cold rolling (Harlet et al., 1993). The latter factor means that
higher reductions per pass can be produced by warm rolling than by cold rolling due to
the lower plasticity of steel at room temperature. In conventional hot rolling, the finish
rolling temperature is above the Ar3 temperature and in warm rolling finish rolling is
made in the ferrite phase.
In the past years many researchers have worked on low, ultra carbon steels (Sakata et al.,
1997) using warm rolling. The products obtained by ferritic hot rolling can be divided into
two kinds according to the coiling temperature (Mao, 2004). One is a thin gauge soft and
ductile hot rolled strip obtained by high temperature coiling for direct application that could
be considered as a substitute for the conventional cold rolled and annealed sheet, and the
other is a strained thin gauge hot strip gained by low temperature coiling for cold rolling
and annealing, during which recrystallization texture strengthens by accumulating strains
from hot rolling and cold rolling reductions.
Sanchez-Araiza et al (Sanchez-Araiza, 2006) reported on the texture changes in low carbon
steel during recrystallization and established the nucleation and growth mechanisms
applicable to warm-rolled quantitatively.
The requirement necessary for the development of (111) texture is the accumulation of strain
in the matrix and to achieve uniform accumulation of strain through the thickness of sheet
steel, lubricant is required in addition to the optimization of Ti and Nb concentration
(Figure 36) (Sakata et al., 1997). In 1996, Kawasaki Steel Corporation constructed a new hot
strip mill in the Chiba works, where sheet bars are welded between the coil box and the
finish mill to accomplish fully continuous rolling. This endless hot strip mill makes
lubricant rolling practical. The r
m
-value achieved using warm rolling technique is 2.9. This
value is noteworthy since the best value obtained in the conventional process does not
exceed 2.6.
Wang et al (Wang et al., 2007) studied drawability of Ti-stabilized IF steel by finish
rolling the steel at 760
o
C in the ferritic region with lubrication and coiling at 740 and
400
o
C. The evolution of texture for both the high and low temperature coiling is shown
in in Figures 37-40. The optical micrographs of the test steels in hot rolled and high
temperature coiled status, cold rolled as well as annealed one are shown in Figure 37. It
can be seen from Figure 37(a) that after high temperature coiling, the deformed
microstructure vanished and the recrystallization microstructure is characterized by
uniform and equiaxed grains. Figure 37(b) shows that after cold rolling, the grains can
not be discerned and the obvious characteristics of the cold rolled microstructure is the
formation of the in-grain bands denoted by the arrow. As shown in Figure 37(c),
deformed microstructure disappears and there are small and elongated grains after
annealing. In order to obtain more equiaxed grains, the annealing temperature or the
annealing time should be increased.
Recrystallization
170
Fig. 36. Effect of lubricant on texture distribution through the thickness in ferrite rolled sheet
steel; with finish rolling temperature 700
o
C, cold reduction 50%, and annealing time 20s at
850 C (Sakata 16 et al., 1997)
Fig. 37. Optical microstructures of samples in the condition of high temperature coiling: (a)
hot rolled (b) cold rolled and (c) annealed samples (Wang et al., 2007).
Figure 38 shows the =45 ODF sections for high temperature coiled hot band, cold rolled and
annealed sample. It is clear that after ferritic rolling and high temperature coiling, the most
prominent texture intensity is along the -fiber and the maximum is at {111}<112>. Moreover, its
characteristics are the same as that of the annealed texture in the condition of low temperature
coiling, indicating that the hot band after high temperature coiling can be considered as a
substitute for the conventional cold rolled and annealed sheet (Wang et al., 2007).
Fig. 38. =45
o
ODF sections in the condition of high temperature coiling: (a) hot rolled, (b)
cold rolled and (c) annealed samples (Wang et al., 2007).
Physical Metallurgy and Drawability of Extra Deep Drawing and Interstitial Free Steels
171
The optical micrographs of the low temperature coiled hot band, cold rolled as well as
annealed one are shown in Figure 39. It is evident that a completely deformed microstructure
is produced after hot rolling and low temperature coiling. Straighter grain boundaries and
thinner deformation bands form after cold rolling. After annealing, the ferrite grains
recrystallize completely, and small and uniform grains develop. Figure 40 shows 2=45
o
ODFs
of low temperature coiled hot band, cold rolled and annealed samples. The texture of hot band
includes a strong -fiber whose peak is at {001}<110> as well as a weak -fiber whose main
component is {111} <110>. The components in the -fiber intensify and the intensity of
{111}<112> in the fiber changes little after cold rolling. A complete -fiber with the peak at
{111}<112> develops and the components in -fiber weaken evidently after annealing.
Fig. 39. Optical microstructures of samples in the condition of low temperature coiling: (a)
hot rolled (b) cold rolled and (c) annealed samples (Wang et al., 2007).
Fig. 40. =45
o
ODF sections in the condition of low temperature coiling: (a) hot rolled, (b)
cold rolled and (c) annealed samples (Wang et al., 2007).
In a study on the effect of chemical composition and ferritic hot rolling on the formation of
texture in the hot rolled and coiled, cold rolled, and cold rolled and annealed Ti and Ti+Nb
added IF steels, it was reported that high r
m
-values could be obtained at the same intensity
levels of {111}<110> texture component if the grain growth was inhibited by a suitable
addition of microalloying elements. (Tiitto et al., 2004). While as per the conventional
wisdom it is known that grain growth during annealing is beneficial for drawability as it
increases the volume fraction of grains with the {111} texture, leading to higher r
m
-value. It
was claimed that due to the grain refining effect of Nb, a small addition of the element (100
ppm) to the Ti alloyed IF steel increased the r
m
-values in the annealed condition by about
15% at the same {111}<110> intensity levels. Thus, ferritic hot rolling seemed to be beneficial
only if it contributed to the development of a strong intensity of the -fiber texture in
Recrystallization
172
association with a uniform and small grain size during annealing. Further they added, hot
deformation of Nb and Ti alloyed steels at a high temperature (870C) in the ferritic region
led to higher {111}<110> intensities and higher r- values in the annealed condition than hot
deformation at a lower temperature (800C).
The development of the -fibre in annealed steels was linked (Barnett & Jonas, 1997a;
Duggan et al., 2000) to the presence of a high volume fraction of grains containing shear
bands after warm rolling. These bands appear to be the nucleation sites for recrystallised
grains of the desirable orientation. After warm rolling, many of the grains in interstitial-free
(IF) steels contain such shear bands and the steels then exhibit good forming characteristics
after annealing (Barnett, 1998). By contrast, in low carbon steels, the presence of carbon in
solid solution leads to dynamic strain aging during warm rolling and to high positive rate
sensitivities. The latter prevent the formation of high densities of in-grain shear bands,
leading to a lack of nucleation sites for the -fibre during annealing.
Timokhina et al (Timokhina et al., 2004) studied the effect of in grain shear bands on the
volume fraction of favourable -fibre in IF, low carbon (LC) and LC with Cr, P and B added
steels. Shear bands are usually contained within single deformed grains and are tilted by 20
35 with respect to the rolling plane. There are four types of in-grain shear bands: long (5-40
m long--15-40
o
to rolling direction), short (0.5-15 m long), intense short (0.4-7 m, 5-45
o
)
and intense long (continuous wavy lines--15-40
o
to rolling direction) (Barnett & Jonas, 1997b)
as shown in Figure 41. All the grains containing shear bands were characterized by zones of
grain boundary displacement or stepping that provide evidence for local flow along the bands
(Barnett, 1996). The presence of moderate amounts of long shear bands in IF steels was
attributed to the formation of the -fibre after annealing (Barnett & Jonas, 1997a).
Fig. 41. SEM micrographs of (a) long shear bands (b) short shear bands, (c) intense long
shear bands and (d) intense short shear bands. Arrows indicate the displacement zones
(Barnett & Jonas, 1997b).
Physical Metallurgy and Drawability of Extra Deep Drawing and Interstitial Free Steels
173
In their study, Timokhina et al (Timokhina et al., 2004) reheated the sample to 1050
o
C for 30
mins and subsequently warm rolled by 65% in a single pass pilot mill. Temperatures of 640
o
C
and 710
o
C and average strain rates of 30s
-1
were employed followed by quenching. To
establish grain orientation with different sheer bands, the grains were marked with hardness
tester prior to electron back scattered diffraction analysis. The addition of alloying elements
was observed to affect different types of shear band and the formation of -fibre. The IF steel
was characterized by the largest number of grains with long shear bands with -fibre, 65 %
and 67 % in the steels rolled at 640C and 710C respectively. Increasing the amount of carbon
and decreasing the grain size, as in the LC steel, increased the proportion of grains with the
short shear bands. Further, the proportion of grains containing intense long shear bands was
higher in the LC steel than in the IF grade. The addition of carbon perceptibly decreased the
volume fraction of grains with -fibre. The addition of chromium led to the formation of
similar volume fractions of grains with short, long and intense long shear bands at both rolling
temperatures. This led to an increase in the number of grains with beneficial -fibre to 4647%
compared to 2526 % in the LC steel. However, in a subsequent study Pereloma et al
(Pereloma et al., 2004) reported that although the formation of chromium carbides in the
microstructures of the LC-Cr and LC-Cr+P steels removed carbon from solid solution and in
this way slightly increased the fraction of -fiber nuclei formed at shear bands, the effect led to
a reinforcement of the ND component (-fiber) only in the early stages of recrystallisation. The
strength of -fiber gradually deteriorated with the progress of recrystallisation. The reasons
put forward for this undesirable development were i) the retarding effect of the carbides on
the mobility of grain boundaries during growth, ii) the influence of the particles on nucleus
rotation during annealing and iii) the absorption of -fiber nuclei by other components during
grain coalescence and growth. Thus, further study of the texture behaviour during grain
coalescence and growth was required.
The addition of boron suppressed the formation of long, short and intense long shear bands
and assisted in the formation of intense short shear bands that, in turn, decreased the
volume fraction of grains with -fibre. The addition of phosphorus, on the other hand,
increased the long and short shear band frequency and the proportion of grains with -
fibre. In the CrBTi modified steel, the short, long and intense long shear bands were
absent and they were replaced by intense short shear bands.
Jing et al (Jing et al., 2011) analyzed the grain boundary and microtexture characters of a
rephosphorised high strength IF-steels under 700 and 800
o
C warm-rolled temperatures to
observe the effect on deep drawability. It was found that while the samples were rolled at
700C, more -fiber texture components, {111}<112>, {111}<110>, {554}<225>, low angle and
CSL grain boundaries were formed that were beneficial for deep-drawability. However, the
samples that were rolled at 800C, manifested more -fiber texture components and high
angle grain boundaries that led to inferior deep drawing property. The average r-value was
1.32 for the samples rolled at 700C and 1.05 for those that were rolled at 800C, respectively.
Recrystallization texture investigation for IF-Ti and Ti stabilized IF-HS was conducted by
Wang et al (Wang et al., 2006) under ferritic hot rolling and high-temperature coiling.
Comparing with the completely recrystallized textures of the ordinary IF steel, the textures
of the high-strength IF steel were of deformation type. This was attributed to the high
phosphorous content in the high-strength IF steel that prevented recrystallization during the
coiling process. For the ordinary IF steel, the texture components were mainly very weak
Recrystallization
174
rotated cube component {001}<110>at the surface, and partial -fiber with key orientation
{223}<110>orientation and <111>||ND texture at the mid-section and 1/4-section. While,
for the high-strength IF steel, the texture components were orientation (Goss) at the surface
and a sharp -fiber extending from {001}<110> to {223}<110in association with a Weak
<111>||ND texture at the mid-section and 1/4-section.
In another work by Ferry et al. (Ferry et al., 2001) on ultra low carbon steel (0.0036C, 0.03Ni,
0.019Mn, 0.03Al, 0.004Ti and 0.003N) it was found that the hot deformation microstructure had
a strong influence both on the kinetics of recrystallization and texture development during
cold rolling and annealing. In particular, a warm-deformed ferrite microstructure (lower finish
deformation temperature (FDT)) recrystallized most rapidly to produce a strong <111>||ND
recrystallization texture ( -fibre) as it produces as-strained in combination with the
additional strain by cold rolling, which result in rapid recrystallization. while an initial coarse-
grained ferrite microstructure recrystallized most sluggishly to produce a strong {001}<110>
texture due to copious nucleation of grains at shear bands, which is consistent with previous
studies (Hutchinson, 1984; Muljono, 2001). Figure 42 (Ferry et al., 2001) shows
2
=45 ODF
sections in the fully recrystallized cold rolled and annealed samples following hot deformation
at three significant finish rolling temperatures: (a) 920C (fine, equiaxed ferrite), (b) 850C
(coarse ferrite) and (c) 600C (warm deformed ferrite). The maximum intensity of the two most
dominant recrystallization texture components, {001}<110> and {111}//ND, as a function of
FDT are given in Figure 43 (Ferry et al., 2001). It can be seen that the development of the
strongest {111}//ND CRA texture is favoured when FDT is: (i) greater than 870C (which
produces fine-grained ferrite by transformation), and (ii) below, 800C (which also produces
fine-grained ferrite but with an additional true strain of 0.8 prior to cold rolling). Thus, it is
indicated that warm rolling has a significant influence on final texture after cold rolling and
annealing and warm rolling is capable of strengthening the {111}//ND (-fiber)
recrystallization texture that is the favourable texture for drawability in the production of
formable ultra low carbon steel sheets.
Fig. 42.
2
=45 sections in Euler space showing recrystallization textures of the ultra low
carbon steel with FDTs 600C, 850C and 920C (contours: 1, 2, 3...random) (Ferry et al., 2001).
Physical Metallurgy and Drawability of Extra Deep Drawing and Interstitial Free Steels
175
Fig. 43. Effect of FDT on maximum texture intensity of {001}<110> and within 10 around
<11l>||ND ( -fibre) of the annealed ultra low carbon steel (the maximum intensity is taken
from the each calculated ODF) (Ferry et al., 2001).
10. References
Auburn, Ph. & Rocquet P. (1973). Mem. Sci. Rev. Met., Vol. LXX, No. 4, p. 261.
Baker L. J., Daniel S. R & ParkerJ. D. (2002). Mater. Sci. Technol. Vol. 18, No. April, pp. 355-
367.
Banerjee K., Verma A. K. & Venugopalan T.(2008). Metall. and Mater. Trans. A, Vol. 39, No.
June, pp. 1410-1425.
Barnett M. R. & Jonas J. J. (1997). ISIJ Int., Vol. 37 pp. 697-705.
Barnett M. R. & Jonas J. J. (1997). ISIJ Int., Vol. 37, pp. 706-714.
Barnett M. R. & Jonas J. J. (1999).ISIJ Int., Vol. 39 pp. 856-873.
Barnett M. R. (1996).Ph.D. Thesis, McGill University, p. 52.
Barnett M. R. (1998). ISIJ Int., Vol. 38 pp. 78-85.
Barnett M.R. & Kestens L.(1999). ISIJ Int., Vol. 39, pp. 923-929.
Barrett M.R. & Jonas J.J.(1997) , ISIJ Int., Vol. 37, pp. 706-714.
Beranger G., Henry G & Sanz G. 1996.The book of Steel, Springer-Verlag, USA, 2-85206-981-
18 (1994), USA, pp. 935-951.
Cao W. D. , Lu X. P. , Sprecher A. F. & Conrad H. (1990). Mater. Lett., Vol. 9, p. 193.
Cao S. Q. (2005). Grain Boundaries and the Evolution of Texture in Interstitial-free (IF)
Steels, Ph.D. Thesis, Shanghai Jiao Tong University.
Chung J. , Era H. & Shimizu M. (1987). Metall. Trans. A, Vol. 18, No. June , pp. 957-968.
Conrad H. , Guo Z. , Sprecher A. F. (1989). Scripta Metall .Vol. 23, pp. 821-823.
Doherty R.D. (1985). Scr. Metall., Vol. 19, pp. 927-930.
Doherty R.D. , Hughes D.A. , Humphreys F.J. , Jonas J.J., Juul Jensen D., Kassner M.E.,
King W.E., McNelley T.R. McQueen H.J. & Rollett A. D. (1997). Mater. Sci. Eng. A,
Vol. 238, pp. 219274.
Duggan B. J. , Liu G. I. , Ning H. , Tse Y. Y. (2000). Proc. Int. Conf. Thermomechanical
Processing of Steels, IOM Communications Ltd. London, UK, pp. 365 - 371.
El-Kashif E., Asakura K. & Shibata K. (2003). ISIJ Int., Vol. 43, No. 12, pp. 2007-2014.
Recrystallization
176
Ferry M., Yu D. & Chandra T. (2001). ISIJ Int., Vol. 41, No. 8, pp. 876882.
Gupta I. and Bhattacharya D. (1990). Metallurgy Of Vacuum Degassed Steel Products, ed.
R. Pradhan, TMS, Warrendale, PA, TMS, pp 43 72.
Harlet P. , Beco F. , Cantinieaux P. , Bouquegneau D. , Messien P. & Herman J. C. (1993). Int.
Symp. on Low C Steels for the 90s, eds. R. Asfahani & G. Tither, TMS-AIME,
Warrendale, PA, p. 389.
Hatherly M. & Hutchinson W. B. (1979). An introduction to textures in metals, The
Institution of Metallurgists, Monograph 5.
He C.S. , Zhang Y.D., Wang Y.N. , Zhao X. , Zuo L. & Esling C. (2003). Scripta Mater., Vol.
48, pp. 73774.
Hebert V., Louis P. , Zimmer P. & Delaneau P. (1992), CESSID document 92154.
Heckler A. J. & Granzow W. G. (1970). Metall. Trans A, Vol. 1, pp. 2089-94.
Held J. F. (1965). Mechanical Working and Steel Processing IV, ed.D. A. Edgecombe,
American Institute of Mining, Metallurgical and Petroleum Engineers, New York,
p. 3.
Holie S. (2000). Mater. Sci .Technol., Vol 16, pp. 1079-1093.
Hlscher M. , Raabe D. & Lucke K. (1991). Steel Res., Vol. 62, No. 12, pp. 567-75.
Hook R. E. (1990). Metallurgy of vacuum-degassed steel products (ed. R. Pradhan), 1990,
Warrendale, PA, Metallurgical Society AIME, p.263.
Hosford W. F. & Backholen W.A. (1964). Fundamentals of deformation processing, Syracuse,
Press, New York. P. 259.
Hu H. & Goodman S.R. (1970). Metall. Trans., Vol. 1, pp. 3057-64.
Hu H. (1977). Metall. Trans. A, Vol. 8, pp. 1567-75.
Hughes I.F. & Page E.W. (1971). Metall. Trans., Vol. 2, pp. 2067-75.
Hutchinson W. B. (1984). Intl. Mater. Reviews, Vol. 29, No. 1, pp. 25-42.
Hutchinson W.B. & Ushioda K. (1984). Scand. J. Met., Vol. 13, p. 269-284.
Irie T. , Hashiguchi H. , Satoh S. , Konoshi M. , Takahashi K. & Hashimoto M (1981).Trans.
Iron Steel Inst. Jpn., Vol. 21, No. 11, pp. 793-801.
Irie T. , Satoh S. , Yasuda A. & Hashimoto O. (1982). Metallurgy of Continuously Annealed
Sheet Steel, TMS, Warrendale, PA, TMS, pp. 155-171.
Jeong W. C. (2000). Metall. and Mater. Trans. A, Vol. 31, No. April, pp. 1305-1307.
Jing C. , Wang M. , Liu X. , Tan Q. , Wang Z. & Han F. (2011). Mater. Sci. Forum, Vol. 682,
pp. 71-74.
Juntunen P. , Raabe D., Karjalainen P. , Kopio T. & Bolle G. (2001). Metall. & Mater. Trans.
A, Vol. 32, No. Aug, pp. 1989-95.
Kang H. , Garcia C. I. , Chin K. & DeArdo A. J. (2007). ISIJ Int., Vol. 47 No. 3, pp. 486492.
Katoh H. , Takechi H. , Takahashi N. & Abe M. (1985). Int. Conf. on Technology of
Continuously Annealed Cold Rolled Sheet Steels, ed. Pradhan. R, Proc. . TMS-
AIME, Warrendale, PA, USA, pp. 37-60.
Kawasaki K., Senuma T. & Sanagi S. (1991). Processing, Microstructure and Properties of
Microalloyed and other Modern HSLA Steels, ISS, Warrendale, PA, pp. 137 - 144.
Kestens L.., Jonas J. J., Van Houtte P. & Aernoudt E. (1996). Textures and Microstructures,
Vol. 26-27, pp. 321-335.
Kim S., Choi I., Park I. & K. Cho (2005). Mater. Sci. Forum, Vol. 475-479, pp. 475-479.
Kitamura M. , Tsukatani I. & Inoue T.(1994) : ISIJ Int., Vol. 34, No. 1, pp. 115 122.
Klein A. J. & Hitchler, E. W. (1973). Met. Eng. Q., Vol. 13, pp. 2527.
Physical Metallurgy and Drawability of Extra Deep Drawing and Interstitial Free Steels
177
Kojima N. , Mizui N.& Tanioku T. (1993). Sumitomo Search, Vol. 45, No. 5, pp.12 19.
Kozeschnik E. , Pletenev V. , Zolotorevsky N. & Buchmayr B. (1999) , Metall. & Mater Trans.
A, Vol. 30, No. June, pp. 1663-1673.
Krauss G., Wilshynsky D. O. & Matlock D. K. (1991). Interstitial Free Steel Sheet: Processing,
Fabrication and Properties, eds. L. E. Collins and D. L. Baragar, CIM/ICM1,
Ottawa, pp. 1- 14.
Lankford W. T. , Snyder S. C. & Bauscher J. A. (1950). Trans.AS1/I, Vol. 42, pp.11971232.
Lebrun J. L., Maeder G. & Parniere P.(1981) : Proc. 6th Intl. Conf on Texture of Materials,
Vol. 2, Tokyo, The Iron and Steel Institute of Japan, p. 787.
Mao X. (2004). Iron Steel, Vol. 39, No. 5, p. 71.
Martin, J.W., Doherty R. D. , Cantor B. (1997). Stability of Microstructure in Metallic Systems
(2nd edition)., Cambridge University Press, Cambridge.
Matlock D. K., Allan B. J. & Speer J. G. (1998). Proc. Conf. Modern LC and ULC Sheet Steels
for Cold Forming Processing and Properties, ed. W. Bleck, Aachen, Verlag Mainz,
pp. 265 - 276.
Matsudo K. , Osawa K. & Kutihara K. (1984). Technology of Continuously Annealed Cold-
Rolled Sheet Steel, ed. R. Pradhan, TMS-AIME, Michigan, pp. 3-36.
McQueen H.J. , Rollett A.D. (1997). Materials Science and Engineering, Vol.A238, pp. 219
274.
Mendoza R. , Huante J. , Alanis M., Gonzalez-Rivera C. & Juarez-Islas J. A. (2000). Mater.
Sci. Eng A, Vol. 276, pp. 203-209.
Mendoza R. ,Alanis M. , Aramburo G. , Serrania F. & Jurez-Islas J.A. (2004). Mater. Sci.
Eng.A, Vol. 368, pp. 249254.
Meyzaud Y.& Parniere P.(1974). Mem. Sci. Rev. Met., Vol. LXXI, No. 7-8, pp. 423.
Mizui N. , Okamoto A. & Tanioku T.(1990). Proc. LTV/SMI Technology Exchange Meeting,
Ltv Steel/Sumitomo Metal Industries.
Muljono D. , Ferry M. & Dunne D.P.(2001). Mater. Sci. Eng. A, Vol. 303, pp. 9099.
Nishimoto A. , Inagaki J. & Nakaoka K.(1982).Tetsu-to-Hagne, Vol. 68, pp.1404-1410.
Ono S. , Shimomura T., Osawa K. & Matsudo K.(1982). Transaction ISIJ, Vol. 22, pp. 732-
738.
Park Y.B. , Kestens L.& Jonas J.J. (2000). ISIJ Int., Vol. 40, pp. 393-401.
Pereloma E. V. , Timokhina I. B. , Nosenkov A. I. & Jonas J. J. (2004). Metallurgija, Vol. 43,
No. 3, pp. 149-154.
Perera M., Saimoto S. & Boyd D. (1991). Interstitial Free Steel Sheet: Processing, Fabrication
and Properties, eds.. L. E. Collins and D. L. Baragar, ;Ottawa, CIM/ICM, pp. 55 -
64.
Ray R. K. & Jonas J. J. (1990). Int., Mater. Rev., Vol. 35, No. 1, pp. 1-36.
Ray R. K., Jonas J. J. & R. E. Hook (1994). Intl. Mater. Reviews, Vol. 39, No. 4, pp. 129-172.
Rege J. S. , . Garcia C. I & DeArdo A. J.(1997). Proc. 39th Mechanical Working and Steel
Processing, Vol. 35, ISS, Warrendale, PA, USA, pp. 149-158.
Rege J. S., Hua C., Garcia I. & DeArdo A. J. (2000). ISIJ Intl., Vol. 40, No. 2, pp.191-199.
Ruiz-Aparicio L.J. , Garcia C.I. & Deardo A.J. (2001) , Metall.and Mater. Trans. A, Vol. 32,
No. September, pp. 2325-2334.
Sanchez-Araiza M. , Godet S. , Jacques P.J. & Jonas J.J. (2006). Acta Mater. , Vol. 54, pp.
30853093.
Recrystallization
178
Sakata K. , Matsuoka S. , Obara T. , Tsunoyama K. & Shiraishi M.(1997). Materia. Japan, Vol.
36 No. 4 p. 376.
Samajdar I. (1994).Ph.D. Thesis, Drexel University.
Sarkar B., Jha B. K. & Deva A. (2004). J. Mater. Eng. and Perform., Vol. 13, No. 3, pp. 361-36
Schulz L. G. (1949). J. Appl. Phys. Vol. 20, No.11, pp.1030-33.
Song X., Yuan Z. Jia J. , Wang D. , Li P. & Deng Z (2010). J. Mater. Sci .Technol. Vol. 26,
No. 9, pp.793-797.
Storojeva L. , Escher C. , Bode R. , K. Hulka & Yakubovsky O. (2000). IF Steels 2000, ISS,
Warrendale, PA, p. 289.
http://www.cbmm.com.br/portug/sources/techlib/report/novos/pdfs/stabiliza.pdf
Takahashi M., Okamoto, A. (1974). Sumimoto Met., Vol. 27,pp. 40-49.
Tanioku T. , Hobah Y. , Okamoto A. & N. Mizui (1991). SAE Technical Paper 910293, Society
of Automotive Engineers, Warrendale, PA, USA.
Tiitto K. M. , Jung C. , Wray P. , Garcia C. I. & DeArdo A. J. (2004). ISIJ Int., Vol. 44, No. 2,
pp. 404413.
Timokhina I. B. , Nosenkov A. I. , Humphreys A. O. , J. J. Jonas & Pereloma E. V.(2004). ISIJ
Int., Vol. 44, No. 4, pp. 717724.
Tither G. & Stuart H. (1995). HSLA Steels '95', ed. L. Guoxun et al.., Chinese Society for
Metals, Beijing, pp. 22-31.
Tither G., Garcia C. I., Hua M. & Deardo A. J. (1994). Int. Forum for. Physical Metallurgy in
IF Steels, Iron and Steel Institute of Japan, Tokyo, pp. 293-322.
Tokunaga Y. & Yamada M. (1985). Method for the Production of Cold Rolled Steel Sheet
Having Super Deep Drawability, US Patent 4,504,326.
Tokunaga Y. & Kato H. (1990). Metallurgy of Vacuum Degassed Products , TMS,
Warrendale, PA, pp. 91 -108.
Tsunoyama K. (1998). Phys. Stat. Sol. (A), Vol. 167, No. 427, pp. 427-433.
Tsunoyama K. , Sakata K., Obara T. , Satoh S. , Hashiguchi K. & Irie T.(1988). Hot and Cold
Rolled Sheet Steels, eds. R. Pradhan and G. Ludkovsky, TMS, Warrendale, PA, pp.
155 - 165.
Tsunoyama K., Satoh S., Yamasaki Y. & Abe H. (1990). Metallurgy of Vacuum Degassed
Products, 1990, TMS, Warrendale, PA, pp. 127 -141.
Wang Y. N. , He C. S. , Zhao X. , Zuo L., Zhi Q. Z. & Liang Z. D (2000). Acta Metall Sinica,
Vol. 36, No. 2, p. A126.
Wang Z. , Guo Y. , Xue W., Liu X. & Wang G.(2007). J. Mater. Sci. Technol., Vol. 23 No.3, pp.
337-341.
Wang Z.D. , Guo Y.H. , Sun D.Q. , Liu X. H. & Wang G.D. (2006). Mater. Charact., Vol. 57,
No. 4-5, pp. 402-407.
Whiteley R. L. & Wise D. E (1962). Flat rolled products III, Interscience, New York, pp. 4763.
Wilshynsky-Dresler D. O., Matlock D. K. & Krauss G. (1995): ISS Mech. Work. Steel Process.
Conf., 1995, 33, pp. 927 940.
Wilson D. V. (1966). J. Inst. Met., Vol. 94, pp. 8493.
Xiaojun G. & Xianjin W. (1995). Textures and Microstructures, Vol. 23, pp. 21-27.
Yasuhara E., Sakata K. , Furukimi O. & Mega T. (1996): Proc. 38th Mechanical Working and
Steel Processing, Vol,. 34, Cleveland, Ohio, USA, pp. 409-415.
Yoshicla K. at. (1974): Deep Drawing Research Group, Proc. 8th Biennial IDDIG Congr.
Gothenburg, 1974, pp. 258268.
8
The Failure Mechanism of Recrystallization
Assisted Cracking of Solder Interconnections
Toni T. Mattila and Jorma K. Kivilahti
Aalto University
Finland
1. Introduction
The typical user environment load spectrum varies significantly between different electronic
applications but changes in temperature are involved in nearly all of them. Owing to the
increasing number of integrated high-performance functions, smartphones and handheld
computers, for example, can experience significant changes in temperature during normal
operation. The changes in temperature are typically caused either by internally generated
heat dissipation or changes in the temperature of the environment. Today the maximum
temperatures inside modern portable electronic products are in the range of 60-70 C but
occasionally they can even rise above 90 C [1,2].
The thermomechanical reliability of electronic component boards has been one of the most
studied aspects in the field for several decades. Sustained interest in this topic has endured
primarily because: (a) the power densities and heat dissipation of novel electronic devices
have increased; (b) electronic devices are being designed for use in ever harsher
environments, such as the engine compartments of automobiles, and (c) newly developed
materials whose long-term behavior in electronic applications is still unknown are
continuously being introduced into electronic assemblies.
Thermomechanical strains and stresses in electronic assemblies are produced when the
thermal expansion and contraction of materials is restricted. The standard thermal cycling
tests extend the temperature range of electronic devices under normal operating conditions
in order to accelerate the accumulation of failures. The maximum extreme temperatures in
some of the standards are set to -65 C and +150 C, but the conditions of -45 C and +125 C
with 15- to 30-minute dwell times are most commonly used [3-5]. The coefficients of thermal
expansion (CTE) of most printed wiring boards (PWB) are much higher than those of most
packages. For instance, the CTE of the most commonly used PWB base material, FR4, is
about 16-17 x 10
-6
/C [6], whereas that of silicon is only 2.5 x 10
-6
/C [7]. Furthermore,
because of the large volume fraction of silicon in electronic packages (see Fig. 1a), packages
have much higher rigidity than PWBs and, consequently, as the component boards are
exposed to changes in temperature, strains and stresses are concentrated in the solder
interconnections between the packages and the PWB, as the natural expansion/contraction
of the PWB is restricted by the packages (see Fig. 1b). Therefore, the reliability of most
electronic products under changing thermal conditions is determined by the ability of the
solder interconnections to withstand thermomechanical loads. Thermomechanical strains
Recrystallization
180
Fig. 1. a) A Ball Grid Array (BGA) package commonly used in high-density electronic
devices; b) structure of the component board assembly; c) formation of strains and stresses
under an increase in temperature.
(a)
(b)
(c)
The Failure Mechanism of Recrystallization Assisted Cracking of Solder Interconnections
181
and stresses in solder interconnections are built up to different extents, of course, at all
length scales, ranging from the submicron intermetallic particles submerged in the tin
matrix of the tin-rich solder interconnections to the structural dimensions of a product.
However, as the CTEs of most printed wiring boards is many times higher than that of most
packages, the influence of strains and stresses at higher length scales (i.e. the board or the
product level) are more influential.
The mechanical properties of solder interconnections are, of course, dependent on the
microstructures formed during soldering but the fact that they are not stable and tend to
change distinctly during the operation of products makes the evaluation of reliability a
challenging task. Thus, detailed understanding of the evolution of microstructures under
thermal cycling conditions and their influences on the failure of solder interconnections is
essential because it can provide us with the means not only to improve reliability but also to
develop more efficient and meaningful methods for the reliability evaluation of and lifetime
prediction.
Despite it being a popular topic of academic research for decades, we are only beginning to
understand the complexities related to the failure mechanisms of solder interconnections
under cyclic thermomechanical loading. Justification for this perhaps surprising statement lies
in the fact that it was only recently that the widely used tin-lead solder alloys were replaced
with new lead-free materials and this change has made a comprehensive re-assessment of
reliability necessary [8,9]. It is well known that the reliability of solder interconnections is
strongly influenced by the microstructures formed during soldering and their evolution
during use, but it is particularly interesting to observe that the microstructural changes in tin-
rich lead-free solders are markedly different from those observed in tin-lead solders, where
failure takes place as a result of the heterogeneous coarsening of tin and lead phases and
eventual cracking of the bulk solder (see, e.g., [10,11]). The microstructural observations of
failed tin-silver-copper solder interconnections have indicated that the microstructures of
solder interconnections change distinctly before cracking but by a different mechanism,
namely cracking that is assisted by recrystallization (see, e.g. [12,15]).
In this chapter we discuss the failure mechanisms of tin-rich lead-free solder
interconnections of electronic component boards from the perspective of the evolution of
microstructures. The focus is placed on the identification of the factors driving the
microstructural evolution in lead-free interconnections that creates the preconditions for the
energetically feasible intergranular propagation of cracks through the interconnections. The
microstrucutural approach to the reliability of electronics is useful, particularly because
many failure mechanisms are related to the inevitable evolution of microstructures that
takes place during the normal operation of products. Owing to the extensive research
carried out over the last decade we know the microstructures and mechanical properties of
many lead-free solder compositions well, but the evolution of microstructures during the
operation of products has still gained little attention. However, before going into details of
the changes in microstructures one should have a generalized understanding of the
microstructures to be discussed when they are in an as-solidified state. Therefore we will
begin this chapter with a brief overview of the solidification and microstructures of tin-rich
lead-free solder interconnections formed during solidification. After that the restoration of
solder interconnections and the conditions under which the recrystallization is initiated are
discussed. The onset and progress of recrystallization in solder interconnections will be
Recrystallization
182
discussed in detail. The failure of the solder interconnections is discussed in a separate
section to emphasize the fact that the propagation of cracks takes place after the change in
microstructures as a result of recrystallization and that recrystallization significantly
enhances their propagation in the solder interconnections. Finally, a method to predict the
changes in microstructure and, ultimately, to predict the lifetime of solder interconnections
is presented and discussed.
2. As-solidified microstructures of tin-rich solder interconnections
The majority of lead-free solders are based on tin (Sn), with a few alloying elements.
Silver (Ag) and copper (Cu) are the most common major alloying elements but alloys
with minor additions of elements such as nickel, antimony, indium, germanium,
manganese, bismuth, zinc, or rare earth elements are also commercially available.
However, for the sake of simplicity let us consider the three-component SnAgCu alloy
with near-eutectic composition, which is the most commonly used solder alloy in the
electronics industry.
The eutectic composition of the tin-silver-copper alloy is about Sn3.4Ag0.8Cu [13,14]. Nearly
all of the solder compositions used in the electronics industry, such as the composition of
Sg3.0Ag0.5Cu that is most widely used today, have silver and copper concentrations below
the eutectic concentration. For such compositions, the tin-rich phase (i.e., high-tin solid
solution) is formed first at the beginning of solidification and its morphology strongly
affects the solidified microstructure. Owing to the high tin content of the near-eutectic
SnAgCu alloys (more than 95 wt-% Sn), the solidification and generated microstructures of
the interconnections are most significantly influenced by the solidification of the tin-rich
phase, even though in practice the tin-rich phase may or may not always be the first phase
to form during solidification. This is because the dissolution of the contact metallizations
changes the nominal composition of the solder and can thereby influence solidification, as
will be pointed out shortly.
Figures 2a-b show an example of a cross-section of a near-eutectic SnAgCu solder
interconnection in the as-solidified condition as imaged by employing the optical (cross-
polarized) microscopy and the scanning electron microscopy. The boundaries between the
contrasting areas in Figure 2b are high-angle boundaries between the matrices of
solidification colonies
1
(the orientation difference between adjacent regions is quite large,
larger than about 15). A uniformly oriented cellular solidification structure of tin is
enclosed within the colony boundaries. There is a low-angle orientation difference between
the cells enclosed by the high-angle colony boundaries. The cellular structure of tin is clearly
distinguishable as cells are surrounded by eutectic regions (see Fig. 2c-d). It is also
noticeable that the solder interconnections, such as that shown in Figure 2, are commonly
composed of a few solidification colonies of relatively uniformly oriented tin cells. [12,15]
Similar observation have been made by other groups also [16,17-21].
1
We have chosen to use the term solidification colony in order to emphasize the fact that under the
reflow conditions employed, a cellular structure is generated in which the difference in crystal
orientations between individual cells is small (few degrees or less). The use of this term also helps us to
make a verbal distinction between the as-solidified microstructures and the recrystallized grains.
The Failure Mechanism of Recrystallization Assisted Cracking of Solder Interconnections
183
Fig. 2. The as-reflowed microstructures of a near-eutectic SnAgCu interconnection: (a)
optical bright field image of a cross-section; (b) a cross-polarized light image of the same
cross-section highlights the colony boundaries (high angle boundaries); (c) an SEM
micrograph showing the cellular structure within a colony, where the cells are separated by
low-angle boundaries; (d) the cell structure of a solidification colony is emphasized by the
small intermetallic particles that surround the tin cells (the sample has been selectively
etched).
Dendritic morphologies of the tin-rich phase are also reported in the literature; see e.g. [22-
25]. It is evident that this phase can solidify in different morphologies depending on the
solidification conditions (e.g., cooling rate and metallizations in contact with the solidifying
solder) and the nominal composition of the solder. The volume of the solder
interconnections obviously influences the microstructures formed during solidification, as a
dendritic structure of the tin-rich phase is more often observed in the case of large solder
volumes, such as cast dog-bone or lap-joint pull test specimens (prepared for the mechanical
characterization of solders) or packages with a relatively large ( 1 mm) bump diameter (e.g.
[16,26-28]). However, the as-solidified microstructure in the near-eutectic solder
interconnections in our studies has consistently been cellular, regardless of the compositions
of the near-eutectic SnAgCu paste or bump alloy, contact pad metallization, package type or
dimensions, or the setup parameters of our full-scale forced convection reflow soldering
oven.
Recrystallization
184
In order to rationalize the formation of the observed microstructures, it is useful to examine
the solidification of solder interconnections with the help of equilibrium phase diagrams. It
should be kept in mind, however, that the equilibrium diagrams do not contain information
about either the effect of the cooling rate or the morphology of the phases and, thus, the
solidification structures are examined as equilibrium solidification
2
. Figure 3 presents the
tin-rich corner of the SnAgCu phase diagram with the isothermal lines representing the
liquidus temperatures, and the primary phase regions of the tin-rich phase, Cu
6
Sn
5
and
Ag
3
Sn.
Fig. 3. Tin-rich corner of the SnAgCu phase diagram with isothermal lines and primary
phase regions [29]. The arrow between the two circles represents the change in nominal
composition (circle) owing to copper dissolution.
2
For example, in order to take the undercooling (which will be discussed shortly) into consideration,
one should extrapolate the liquidus surfaces of the Cu6Sn5 and Ag3Sn phase to lower temperatures and
lower (or remove) the liquidus surface of the (Sn)prim. phase.
The Failure Mechanism of Recrystallization Assisted Cracking of Solder Interconnections
185
Let us consider, for example, the composition of Sn3.0Ag0.5Cu, which is a very commonly
used alloy in the reflow soldering of component boards. The solidification of this composition
starts with the formation of the tin-rich solution phase when the interconnections are cooled
down from the peak reflow temperature to below the liquidus temperature of about 220 C.
The secondary phase, namely Cu
6
Sn
5
or Ag
3
Sn, is formed only after the nominal composition
of the remaining liquid meets the curve of two-fold saturation, after which the solidification of
the interconnections proceeds by the binary eutectic reaction (liquid transforms to (Sn)
eut
+ the
secondary phase; see [15] for more details).
It should also be noticed that in practice the microstructures formed on other pad
metallizations can differ notably, even though the same solder compositions are used (see,
e.g., [15,30,31]). The dissolution of the contact pads or pad metallizations of packages and
printed wiring boards during reflow changes the composition of the molten solder. The
influence of contact metallizations depends primarily on the dissolution rate and reactivity
of the metallizations. For example, the dissolution rate of copper in near-eutectic SnAgCu
solder is about 0.07 m/s, but that of Ni is more than an order of magnitude lower and can
be considered negligible [15,32,33]. Thus, practically all nickel that is dissolved into liquid
solder is used in the reaction to form intermetallic layers. However, in the cases where the
solder is in direct contact with copper pads (i.e., no protective coating or organic soldering
preservative is used on the copper soldering pads) the dissolution rate of copper from the
pads into typical BGA solder interconnections (with a bump diameter of about 0.5 mm) is
high enough to lift the copper concentration above 1 wt-% during soldering (see Fig. 3). This
change in the nominal composition of the liquid can change the primary phase formed
during solidification from the tin-rich phase to Cu
6
Sn
5
. Therefore, the as-solidified
microstructures on the copper pads often show large amounts of primary Cu
6
Sn
5
(hexagonal) tubes or rods in the microstructure that are absent from the microstructures of
interconnections that are soldered on slow dissolution rate metallizations, such as nickel.
Furthermore, interconnections soldered on copper pads, as opposed to those soldered on
nickel, typically show more numerous and larger Cu
6
Sn
5
particles embedded at the
boundaries between the tin cells that are formed in the binary solidification, as the
composition of the liquid moves a greater distance along the eutectic valley (along the curve
of two-fold saturation).
There is an important consequence related to the increased copper content: the relatively
large primary phase needles or particles dispersed in the solder interconnections can
influence the evolution of solder interconnection microstructures. The non-coherent high-
angle boundaries between the Cu
6
Sn
5
crystals and tin matrix provide good nucleation sites
for the recrystallization. It has been previously demonstrated that the second phase particles
can accelerate the nucleation of recrystallization in common structural alloys [34,35].
Readers interested in particle-stimulated nucleation of recrystallization can refer e.g. to
[36,37].
However, as pointed out earlier, in practice the solidification process departs somewhat
from that of equilibrium solidification. Taking the undercooling into account would result in
the nucleation of the Cu
6
Sn
5
as a primary phase at even lower copper concentrations. Figure
4 illustrates an as-solidified microstructure of the Sn3.0Ag0.5Cu solder interconnection. It is
interesting to observe that the primary Cu
6
Sn
5
needles in the micrographs have the tendency
to nucleate at the free surfaces of the molten interconnection, most probably on oxide
Recrystallization
186
Fig. 4. As-solidified microstructure of a Sn3.0Ag0.5Cu solder interconnection illustrate how
the primary Cu
6
Sn
5
particles have nucleated on oxide particles on the surface of the liquid.
The Failure Mechanism of Recrystallization Assisted Cracking of Solder Interconnections
187
particles on the liquid surfaces, instead of the package or the PWB side interfaces. It should
also be mentioned, without going into detail, that it seems as if the solidification of the tin-
rich phase is controlled by the kinetics of heterogeneous nucleation at the surface of (Cu
6
Sn
5
)
intermetallic layers [38]. Darveaux et al. observed experimentally that SnAgCu solder alloys
with a higher copper concentration exhibit a higher amount of undercooling than those with
a lower concentration [39].
Before we move on to the evolution of microstructures under operating conditions, we
would like to point out a few aspects to consider in more detailed investigations of the
failure mechanisms of recrystallization-assisted cracking of solder interconnections. It is
particularly noteworthy that the Cu
6
Sn
5
or the Ag
3
Sn phase can nucleate with minimum
undercooling in the liquid SnAgCu interconnections [40-42] but the nucleation of the tin-rich
phase results in a significantly wide range of undercooling that can extend up to 60 C [43-
48]. The large amounts of undercooling indicate apparent difficulties in the nucleation of tin
crystals in the liquid, which can be one of the reasons why there are very often only few
orientations of the Sn-rich phase observed on a cross-section of solder interconnections. The
tendency to form only a few large crystals, which can be several hundred micrometers in
diameter, has been observed in interconnections of various length scales, ranging from
about 100 m (the diameter of a Flip Chip interconnection) to millimetre scale (lap-joint
specimens used in material characterization) [15-20,28,49]. Furthermore, the fact that
sometimes neighboring regions of a cross-section share a twinning relationship (indicated
by a misorientation angle of about 60 between adjoining regions) suggests that these
crystals originate from a common nucleus and, thus, the number of different crystals can be
even smaller that the amount determined by the commonly employed qualitative method of
cross-polarized light microscopy [16,20,21].
What has been stated above indicates clearly that mechanical behavior of solder
interconnections is most probably quite different from that of a normal polycrystalline
material. As has been pointed out and is currently being studied by many authors, the fact
that the physical and mechanical properties of the tin-rich phase exhibit significant
anisotropic behaviour
3
can cause severely non-homogeneous deformation and the formation
of internal stresses in the solder interconnections [52-54]. As the internally produced strains
and stresses are combined with the higher-level strains and stresses, as caused by the
differences in the coefficients of thermal expansion of printed wiring boards and packages, it
is clear that the thermomechanical response of the solder interconnections becomes very
complex and unique to each solder interconnection. Furthermore, the grain boundary
cracking should not occur in the as-solidified structure due to the absent of high angle
boundaries (other than those between colonies). This can lead to unpredictable failure sites,
as reported in [54,55]. Therefore, when stress is applied to interconnections having this kind
of microstructure, they undergo microstructural evolution before fractures can propagate.
Investigations of the microstructures of failed solder interconnections have indicated that
the microstructures formed during solidification are not stable and will change notably
during the operation of products [15,17,18,56-63].
3
The coefficient of thermal expansion along the c-axis of the tetragonal unit cell {c/a ratio of about 0.5}
is twice that along the other two axes {a and b-axis}; the elastic modulus along the c-axis is only about
0.6 times that along the other two axes [[50],[51]]
Recrystallization
188
3. The role of recovery and recrystallization in the failure mechanisms of
solder interconnections
The reliability of electronic devices is commonly assessed by employing standard thermal
cycling tests that place the extreme temperatures in the range of about - 45 C to +125 C.
The thermomechanical stresses formed in the solder interconnections under these conditions
are high enough to cause instantaneous plastic deformation of the commonly used near-
eutectic SnAgCu solders [63]. Furthermore, this whole temperature range remains above the
0.3-0.4 homologous
4
temperature range of the solders, which is the temperature range above
which the time-dependent deformation of metals becomes significant. Thus, time spent at
either elevated or lowered temperatures allows diffusion creep processes to transform the
elastic strain part of the total strain into inelastic strain
5
. The energy stored during
deformation acts as the driving force for the evolution of the microstructures.
The initiation of microstructural changes in solder interconnections is localized because of
the highly non-uniform distribution of strains inside the solder interconnections. Figure 5
shows the calculated strain energy density distribution in the cornermost solder
interconnection of the component board assembly shown in Figure 1. It should be pointed
out that there is a difference in the rate of strain energy accumulation during thermal cycling
between the solder regions on the opposite sides of the interconnections and, therefore,
changes in the microstructure are observed first on the package side regions of the
interconnections, where the inelastic deformation is more extensive.
Fig. 5. Calculated strain energy density of the cornermost solder interconnections of the
component board assembly sketched in Figure 1 [63,64].
4
Defined as the ratio of the prevailing temperature to the melting point of a solvent metal; both
expressed in Kelvin.
5
Here we consider total strain = elastic + inelastic strain = elastic + plastic + creep strain.
The Failure Mechanism of Recrystallization Assisted Cracking of Solder Interconnections
189
In studies reported in more detail elsewhere, the evolution of microstructures was
investigated as a function of thermal cycles by taking out samples at fixed intervals
during the course of the test and inspecting them for the development of microstructures
and failures [64,65]. The results showed that the evolution of microstructures in the
strain concentration regions commenced with gradual evolution of the cellular
solidification structure, but after some time, i.e., the incubation period, the
microstructures changed discontinuously by recrystallization. A similar observation has
been reported in [66].
3.1 Restoration of tin-rich solder alloys
When solder interconnections are deformed plastically, a part of the work is stored in
solders as lattice defects, mainly in the strain fields of dislocations. The increased internal
energy of deformed solder acts as the driving force for the competing restoration processes,
recovery and recrystallization. It is well known that the degree of restoration by recovery
depends on the stacking fault energy of the solder alloy. At the time of writing there is little
information in the literature about the recrystallization behavior of tin-based solder alloys
but since the near-eutectic SnAgCu alloy contains more than 95 wt-% of tin, recrystallization
studies on pure tin can be considered indicative, bearing in mind that the alloying elements
in solid solutions, as well as small precipitates, do affect the restoration processes. Creep
studies carried out with high-purity tin have suggested that the stacking fault energy of tin
is high [67,68]. The recovery is very effective in high-stacking fault energy metals, such as
aluminum and iron, as a result of the efficient annihilation of dislocations by cross slip and
climb. Therefore one can expect the restoration of high-tin solder alloys to take place to a
large extent by recovery. Gay et al. and Guy have observed that pure tin (99.995% purity)
recrystallizes at room temperature even after a modest deformation (reductions of a few
percent) [69,70]. However, in a more recent study Miettinen concluded that that even highly
deformed (up to a 50% reduction) near-eutectic SnAgCu solders do not recrystallize
statically when annealed at 100 C after deformation at room temperature [71]. Korhonen et
al. also failed to observe recrystallization in dynamic fatigue tests performed at room
temperature [72]. All these results indicate that recovery is effective also in near-eutectic
SnAgCu solder alloys. Because recovery and recrystallization are competing processes, the
progress of recovery can reduce the driving force of recrystallization significantly and
recrystallization may not always initiate. On the other hand, it is well documented that near-
eutectic SnAgCu interconnections do recrystallize under dynamic loading caused by
changes in temperature (between -45 C and +125 C), as well as under power cycling
conditions (between room temperature and +125 C) [15,17,18,56-63,73]. Thus, it seems that
near-eutectic SnAgCu solder interconnections recrystallize only under restricted loading
conditions: dynamic loading conditions where the strain hardening is more effective than
the recovery.
Figure 6a shows a micrograph of the recrystallized microstructure on the package side neck
region of a thermally cycled interconnection taken by employing optical microscopy and
cross-polarized light. Figure 6b shows an electron backscatter diffraction orientation map of
the same surface. The black lines in the figure represent the boundaries where the crystal
orientation of the adjacent grains exceeds 30 and correspond well to the grain boundaries
visible in the optical micrograph in Figure 6a.
Recrystallization
190
Fig. 6. a) Optical micrograph showing the recrystallized structure on the package side
interfacial region of the SnAgCu solder interconnection taken with polarized light;
b) EBSD graph of the same location as in a) showing boundaries with large misorientation
(larger than 30C) between the adjacent grains with black lines.
3.2 Early phase of evolution: Effects of recovery and coarsening of the
microstructures
Figure 7 shows a collage of micrographs that are all taken from the same solder
interconnection. It should be noticed that we are using the same interconnection to
exemplify the features of microstructural evolution that take place consequently in the stress
concentration regions of the solder interconnections. The evolution of microstructures on the
PWB side interfacial region is much slower than the evolution on the package side of the
interconnection as a result of the less extensive plastic deformation per cycle (see Fig. 5).
Even though this interconnection has experienced more than 3000 cycles and failed from the
package side interfacial region, the microstructures visible on the PWB side of the
interconnections are very similar to those on the package side interfacial regions about 1000
to 1500 cycles earlier.
As already discussed, the as-solidified microstructures of tin-rich solder interconnections are
typically composed of relatively few large tin colonies distinguished by high-angle
boundaries. The cellular structure of the tin-rich colonies is clearly visible within the high-
angle boundaries as the individual tin cells are surrounded by eutectic regions composed of
fine Cu
6
Sn
5
and Ag
3
Sn particles. As shown in Figure 7b the cellular structure of the primary
tin-rich phase is still visible in the less deformed regions on the solder interconnections, such
as in the middle, even after the component board has been thermally cycled until failure.
However, in the strain concentration regions the tin cells begin to rearrange by the gradual
coalescence of the tin cells and coarsening of the intermetallic particles (see Fig. 7c-d),
during which the eutectic structures around the tin cells gradually disappear and a network
The Failure Mechanism of Recrystallization Assisted Cracking of Solder Interconnections
191
Fig. 7. a) Cross-polarized light micrograph of a solder interconnection that has failed under
thermal cycling; b-d) magnifications from the regions indicated in (a); e) EBSD map of the
cross-section that shows boundaries with a misorientation < 15 by yellow lines, 15-45 by
green lines, and > 45 by red lines; f) histogram distribution of grain boundaries over the
region shown in (e). [64,65]
Recrystallization
192
of low-angle boundaries produced by recovery emerges. Figure 7e shows an EBSD map of
the PWB side interfacial region of the same cross-section as shown in Figure 7a. The yellow
lines represent the low-angle boundaries, i.e., boundaries where the crystal orientation
between the adjacent grains is well below 15 (see also Fig. 7f). The green and red lines
represent high-angle boundaries where the misorientation between adjacent grains is higher
than 15, the green line those between 15 and 45, and the red line those above 45. The
figure shows that in addition to the coarsening of the intermetallic particles, the formation of
additional low-angle boundaries (additional to those between the cells of the Sn-rich phase
formed during solidification) takes place in the regions of high strain energy density. These
changes were observed to initiate early in the course of the thermal cycling tests, within
about 50 to few hundred thermal cycles. It is also noteworthy that the coarsening of the
intermetallic particles is strong in the regions near the high-angle grain boundaries, while
the regions near the low-angle boundaries still include finer particles, comparable to the
bulk of the solder ball (see Fig. 7d-e). This can be expected as the diffusion is much faster
along the high-angle boundaries than it is along the low-angle boundaries.
3.3 Later phase of evolution: Transformation of the microstructure by recrystallization
In the course of further thermal cycling, the microstructures keep evolving gradually in the
manner already described until these regions change discontinuous into more or less
equiaxed grain structures by recrystallization. This change in the microstructures is first
observed in the edge regions (i.e. corner regions of a cross-section) of the solder
interconnections on the package side of the interconnections (the regions with the highest
strain; see Fig.6) and, after the initiation of recrystallization, the recrystallized volume
gradually expands from the edges toward the center, across the interconnections near the
package side interfacial region of the interconnections. The incubation time of
recrystallization varies significantly from one interconnection to another and even from one
package to another (the same location of the interconnection) under the same loading
conditions. The first indications of recrystallization in the BGA packaged board assemblies
shown in Figure 1 were observed after about 500 thermal cycles but it can take up to about
2000 cycles until recrystallization is consistently observed in every interconnection in the
corner regions of the packages.
Figure 8 shows a typical example of a failed interconnection, where the cracking of the
solder interconnection is accompanied by a distinct change in the microstructure by
recrystallization. The micrograph in Figure 8a is an optical micrograph that shows the crack
path distinctly and the micrograph in Figure 8b is a cross-polarized light image of the same
location as Figure 8a. The comparison of the micrographs shows that the propagation path
of the crack is enclosed entirely within the recrystallized region of the interconnection.
Figure 8c shows an EBSD map of the same cross-section as shown in Figures 8a-b. The
colour lines represent the misorientation between the adjoining regions: yellow below 15,
green between 15 and 45, and red above 45. The image illustrates well the fact that the
cracked region on the package side interface of the solder interconnections shows primarily
high misorientations and that the high-angle grain boundaries are located very close to the
crack path while, in general, the misorientations become smaller with increasing distance
from the crack region. Figure 8d shows a histogram distribution of grain boundaries with
different orientation over the region shown in Figure 8c (see also Fig 7f). As can be seen, the
region still contains a high number of low-angle boundaries (caused by recovery) but a large
The Failure Mechanism of Recrystallization Assisted Cracking of Solder Interconnections
193
Fig. 8. a) an optical micrograph of a failed solder interconnection shows the crack path
clearly (the thin black line between the gray intermetallic layer and the white solder is a
contrast effect caused by specimen preparation); b) a cross-polarized light image of the same
location as (a) highlights the recrystallized grains as caused by the cyclic deformation; c)
EBSD map of the cross-section that shows boundaries with misorientation < 15 by yellow
lines, 15-45 by green lines, and > 45 by red lines; d) histogram distribution of grain
boundaries over the region shown in (c). [64,65]
Recrystallization
194
number of higher-angle boundaries (caused by recrystallization) have emerged. After the
initiation of recrystallization in the strain concentration regions at the edges of the solder
interconnections, the recrystallized volume gradually expands over the diameter of the
interconnections, and cracks follow the expansion of the microstructurally changed volume.
It was also observed that cracks rarely propagate outside the recrystallized volume of the
solder interconnections.
4. Cracking of recrystallized solder interconnections
Work presented in more detail elsewhere focused on the evaluation of the nucleation time
and propagation rate of cracks in BGA component board assemblies under different thermal
cycling conditions [74,75]. Figure 9 shows the average crack lengths of the most critical
solder interconnections as measured from cross-sections prepared along the diagonal line of
the package as a function of thermal cycles. The different lines represent different cycling
conditions (TS = thermal shock, TC = thermal cycling; the accompanying value is the dwell
time of the profile in minutes).
Fig. 9. Measured average crack lengths as a function of the number of thermal cycles (the
numbers in parentheses are estimates of the Weibull characteristic lifetimes []). [74,75]
More detailed examinations showed that the nucleation of cracks in the interconnections of
the BGA board assemblies took place within a relatively narrow range, between about 1000
and 1500 cycles, regardless of the dwell time or ramp rate used in the thermal cycling tests.
However, the propagation rate of cracks without the influence of recrystallization was very
slow. This conclusion was made based on the comparison of the measured crack lengths in
interconnections that were removed from the thermal cycling oven at the same time and that
The Failure Mechanism of Recrystallization Assisted Cracking of Solder Interconnections
195
showed or did not show recrystallization. Thus, the primary failure mechanism under
thermomechanical fatigue involves the formation of a continuous network of grain
boundaries by recrystallization that enables cracks to nucleate and propagate
intergranularly through the solder interconnections (see Fig. 10).
Fig. 10. Cracks propagate intergranularly between the recrystallized grains: a) cross-section
of a recrystallized and cracked solder interconnection; b) the same image as in (a) but with
superimposed crack paths.
The fractographic examinations illustrate the influence of recrystallization on crack
propagation. Figure 11 shows the fractrographs of failed solder interconnections. The
fracture surfaces exhibit a globular appearance as a result of the propagation of cracks
between the recrystallized grains. Fatigue striations were occasionally observed on the
inspected fracture surfaces. Although they were quite uncommon, they indicate that cracks
can also propagate transgranularly under conditions when the intergranular propagation is
not benignant, namely when the recrystallized grain size is larger or the stress state,
orientation, and geometry of the grains are not in favor of cracking along the grain
boundaries. The networks of the grain boundaries formed by recrystallization evidently
provide favorable paths for cracks to propagate intergranularly with less energy
consumption in comparison with transgranular propagation. It can also be expected that the
cohesion between the recrystallized tin grains is lowered by grain segregation of impurities
as well as locally by intermetallic particles. Furthermore, the mechanical anisotropy of the
(recrystallized) tin grains can also enhance the nucleation and propagation of microcracks
along their boundaries as the value of the coefficient of thermal expansion of tin single-
crystal in the [100] = [010] directions is about two times that in the [001] direction [50,51].
Recrystallization
196
a) TS10
b) TS0
c) TC0
Fig. 11. Fractographs of the interconnections viewed from the component side [75]. Note the
several secondary cracks visible in all micrographs (striations are not visible here).
The Failure Mechanism of Recrystallization Assisted Cracking of Solder Interconnections
197
As discussed in the previous chapter, the range of the incubation periods of recrystallization
in tin-rich solder interconnections is relatively large, in the range of 500 to 2000 cycles in the
case of the BGA interconnections in our study. Thus, the nucleation of cracks can take place
before or after the change in microstructures by recrystallization. However, the failure
analyses showed that the formation of the networks of grain boundaries by recrystallization
had influenced the propagation of cracks in all electrically failed interconnections. The range
of 1000-1500 thermal cycles required to produce distinguishable small cracks (nuclei) equals
about 25-30% of the average lifetime of the component board assemblies and, thus, cracks
are in the propagation stage for about three quarters of the lifetime of component boards.
Therefore, it is evident that the cracking of tin-rich solder interconnections is controlled by
the rate of recrystallization.
On the basis of the results presented above we can draw two conclusions: 1) the nucleation
of cracks in solder interconnections is primarily dependent on the number of load reversals.
In other words, nucleation is relatively insensitive of microstructural features and their
evolution, as well as the parameters of the loading conditions; 2) the rate of crack
propagation is dependent on the expansion rate of the recrystallized volume. In other
words, the lifetimes of solder interconnections are primarily controlled by the onset and
expansion of recrystallization.
5. An approach to lifetime prediction based on the evolution of
microstructures
A thorough understanding of the restoration process in solders can allow the development
of methods for improved lifetime estimation that are based on the evolution of
microstructures. Work presented in [76] describes an approach to lifetime prediction based
on the competing nature of the restoration processes: under conditions in which the strain
hardening is more effective than recovery, the cyclic deformation accumulates the stored
energy above a critical value, after which the recrystallization can initiate. The total stored
energy of the system consists of the grain boundary energy and the volume defect energy
(mainly line and point defects). The stored energy is released through the nucleation and
growth of new strain-free grains (grains with low defect density), which gradually consume
the strain-hardened matrix of high defect density. Li et al. have developed a multiscale
model based on this principle for predicting the microstructural changes of recovery,
recrystallization, and grain growth in solder interconnections subjected to dynamic loading
conditions [77,78]. The approach developed in this work is based on the principle that the
stored energy of the solder is gradually increased during each thermal cycle. When a critical
value of the energy is reached, recrystallization is initiated. It is assumed that, even though
recovery consumes a certain amount of the energy, the net change in the energy per cycle is
always positive as experimental investigations have shown that newly recrystallized grains
consistently appear sooner or later under various thermal cycling conditions. The stored
energy is released through the nucleation and growth of new grains, which gradually
consume the strain-hardened matrix of high defect density.
The approach is realized by combining Monte Carlo simulations with finite element
calculations. The Monte Carlo method is employed to model the mesoscale microstructure
and the finite element method to model the macroscale non-homogeneous deformation (see
Recrystallization
198
Fig. 12). The non-homogeneous volume stored energy distribution in solder
interconnections is scaled from the finite element model results and mapped onto the lattice
of the Monte Carlo model. The quantitative prediction of the onset of recrystallization is
carried out with the help of the Monte Carlo simulation.
Fig. 12. Flow chart for the simulation of microstructural changes in solder interconnections
[77,78]. Acronym TC stands for thermal cycling.
In the Monte Carlo lattice, two adjacent sites with different grain orientation numbers are
regarded as being separated by a grain boundary, while a group of sites with the same
orientation number are considered a single grain. The total stored energy of the system
under consideration consists of the grain boundary energy and the volume defect energy.
Each site contributes an amount of stored energy to the system, and each pair of dissimilarly
oriented neighboring sites contributes a unit of grain boundary energy to the system. The
recrystallization process is modeled by randomly introducing nuclei (small embryos with
zero stored energy) into the Monte Carlo lattice at a constant rate. An non-recrystallized site
will become recrystallized if (a) the volume stored energy of the chosen sites is larger than
the critical stored energy, and (b) the total energy of the system is reduced. If the selected
site is recrystallized, it is considered as a contribution to the grain growth process.
The computational results were compared with the experimentally observed
microstructural changes in solder interconnections subjected to thermal cycling tests. The
results of the microstructural simulations carried out in this work can be summarized as
The Failure Mechanism of Recrystallization Assisted Cracking of Solder Interconnections
199
follows: the incubation period of the recrystallization is about 1000 thermal cycles under the
particular cycling conditions. The recrystallization is initiated in the corner regions on the
package side of the solder interconnections and the expansion of the recrystallized region is
controlled by the volume stored energy distribution. The expansion takes place first along
the package side interfacial region toward the center of the interconnections. After that the
recrystallized region expands toward the rest of the interconnections.
The incubation period of the recrystallization, the expansion of the recrystallized region, and
the rate of increase of the recrystallized fraction are in good agreement with the experimental
observations of thermally cycled component boards. This method predicts reasonably well the
incubation period and the growth rate of the recrystallization, as well as the expansion of the
recrystallized region. Figure 13 shows a comparison of predicted microstructural evolution
with experimental evolution. The onset of recrystallization is a useful criterion to determine
when the material models for the as-solidified microstructures are not valid anymore and,
therefore, crack nucleation and propagation should be taken into account.
500 cycles 1000 cycles 1500 cycles 2500 cycles 3000 cycles 5000 cycles 6000 cycles
Fig. 13. Observed and simulated microstructural changes of solder interconnections with an
increasing number of thermal cycles [77].
In a more recent work Li et al. [78] have developed the method to take into consideration the
fact that the nucleation of new grains of low defect density is more likely at the boundaries
of high misorientation, such as the boundary between the intermetallic particles and the tin
matrix. In particular, the coarse primary intermetallic particles can generate localized stress
concentrations under an applied load because of their dissimilar mechanical properties with
respect to the tin matrix. The intermetallic particles are introduced in the Monte Carlo model
as inert particles that also do not move or grow. In practice the fine and uniformly
distributed intermetallic particles would suppress the recrystallization to some extent by
Recrystallization
200
influencing the motion of the grain boundaries of recrystallizing grains but this is not (yet)
considered in the model.
6. Conclusions
The cracking of the near-eutectic SnAgCu interconnections under thermal cycling conditions
occurs through the bulk of the solder interconnections after a change in the microstructure
by recrystallization. The cumulative increase in the stored energy during each deformation
cycle provides the driving force for the recrystallization. After the solidification structure
has changed into a more or less equiaxed grain structure, there is a continuous network of
new high-angle boundaries providing favorable sites for cracks to nucleate and propagate
with less energy consumption in comparison with the cracking of the as-soldered
microstructure. The decrease in the stored energy in near-eutectic SnAgCu solders is
assumed to take place very effectively by the recovery resulting from the high stacking fault
energy of tin. Therefore, the recrystallization is initiated under well-defined loading
conditions: dynamic loading where strain hardening is more effective than recovery.
The recrystallization of solder interconnections under thermomechanical (or in cyclic power)
loading is an important phenomenon for the following reasons: being an experimentally
observable indicator of microstructural evolution, the recrystallization enables one to
establish a correlation between the field use loading conditions and those produced in
accelerated reliability tests. Furthermore, the theoretically well-known phenomena of
recovery and recrystallization can provide the means to incorporate the effects of
microstructural evolution into lifetime prediction models, which are being increasingly
employed to reduce the amount of reliability testing. Finally, by controlling the
recrystallization of solder interconnections, for example by alloying, one may discover new
solutions to improve the reliability of soldered electronic devices.
7. Acknowledgements
The authors wish to thank the following people for their valuable contribution to the work
presented in this chapter: Mr. Jussi Hokka, Dr. Jue Li, Dr. Maik Mueller (TU Dresden,
Germany), Mr. Otso Ratia, and Dr. Hongbo Xu. The authors would also like to thank the
following people for their much appreciated help and the numerous discussions over the
years: Ms. Pirjo Kontio, Ms. Sini Niiranen, Ms. Johanna Koivisto, Ms. Riitta Viitala, Dr.
Hongtao Chen, Dr. Erkki Heikinheimo, Mr. Simo Miettinen, Dr. Tomi Laurila, and Dr. Vesa
Vuorinen. Special thanks go to Prof. Mervi Paulasto-Krckel and Prof. Klaus-Juergen Wolter
(TU Dresden, Germany) for their favorable support for this work. The financial support
from the Academy of Finland (decision number 123922), the Finnish Funding Agency for
Technology and Innovation (decision numbers 40135/07, 662/06, and the ELMO program),
and the Finnish electronics industry is gratefully acknowledged.
8. References
[1] J. Karppinen, T. T. Mattila, and J. K. Kivilahti, Formation of thermomechanical
interconnection stresses in a high-end portable product, The Proceedings of the 2nd
Electronics System Integration Technology Conference, London, UK, September 1-4,
2008, IEEE/EIA CPMT, (2008), pp. 1327-1332.
The Failure Mechanism of Recrystallization Assisted Cracking of Solder Interconnections
201
[2] J. S. Karppinen, J. Li, T. T. Mattila, and M. Paulasto-Krckel, Thermomechanical
reliability characterization of a handheld product in accelerated tests and use
environment, Microelectronics Reliability, (in print).
[3] IEC 60068214 Ed. 5.0 b: 1984, Environmental testing part 2: tests. Test N: change of
temperature, International Electrotechnical Commission, (1984), 34 p.
[4] JESD22-A104C, Temperature Cycling, Jedec Solid State Technology Association,
(2005), 16 p.
[5] IPC-TM-650 rev. A, Thermal Shock and Continuity, Printed Board, The Institute for
Interconnecting and Packaging Electronic Circuits, (1997), 2 p.
[6] C. F. Coombs Jr., Printed Circuits Handbook, 5th ed., New York, (2001), McGraw-Hill,
1200 p.
[7] Y. S. Touloukian and C. Y. Ho, Thermal Expansion: Metallic Elements and Alloys, New
York, (1975), IFI/Plenum, 316 p.
[8] Directive 2002/95/EC of the European Parliament and of the Council on the Restriction
of the Use of Hazardous Substances in Electrical and Electronic Equipment (RoHS),
Jan. 27th, 2003.
[9] Directive 2002/96/EC of the European Parliament and of the Council on Waste of
Electrical and Electronic Equipment (WEEE), Jan. 27th, 2003.
[10] D. R. Frear, Microstructural evolution during the thermomechanical fatigue of solder
joints, in M.J. Cieslak, M.E. Glicksman, S. Kang, and M.E Glicksman, The Metal
Science of Joining, The Minerals, Metals & Materials Society, pp. 191-2000.
[11] J. W. Morris Jr., D. Tribula, T. S. E. Summers, and D. Grivas, The role of microstructure
in thermal fatigue of Pb-Sn solder joints, in John H. Lau, Solder Join Reliability, New
York, 1991, Van Nostrand Reinold, pp. 225-265.
[12] T. T. Mattila, T. Laurila, and J. K. Kivilahti, Metallurgical factors behind the reliability
of high density lead-free interconnections, in E. Suhir, C. P. Wong, and Y. C. Lee,
Micro-and Opto-Electronic Materials and Structures: Physics, Mechanics, Design,
Reliability, Packaging, Springer Publishing Company, New York, 2007, (1), pp. 313-
350.
[13] W. Q. Peng, Lead-free electronic assembly based on SnAgCu solders, Espoo,
licentiate thesis, Helsinki University of Technology, (2001), p. 124.
[14] K.W. Moon, W. J. Boettinger, U. R. Kattner, F. S. Biancaniello, and C. A. Handwerker,
Experimental and thermodynamic assessment of SnAgCu solder alloys, Journal
of Electronic Materials, 29, 10, (2000), pp. 1122-1136.
[15] T. T. Mattila, V. Vuorinen, and J. K. Kivilahti, Impact of printed wiring board coatings
on the reliability of lead-free chip-scale package interconnections, Journal of
Materials Research, 19,11, (2004), pp. 3214-3223.
[16] A. LaLonde, D. Emelander, J. Jeannette, C. Larson, W. Rietz, D. Swenson, and D. W.
Henderson, Quantitative metallography of -Sn dendrites in Sn3.8Ag0.7Cu ball
grid array solder balls via electron backscatter diffraction and polarized light
microscopy, Journal of Electronic Materials, 33, 12, (2004), pp. 1545-1549.
[17] D. Henderson, J. J. Woods, T. A. Gosseling, J. Bartelo, D. E. King, T. M. Korhonen, M. A.
Korhonen, L. P. Lehman, E. J. Cotts, S. K. Kang, P. Lauro, D.-Y. Shih, C. Goldsmith,
and K. J. Puttliz, The microstructure of Sn in near eutectic Sn-Ag-Cu alloy solder
joints and its role in thermomechanical fatigue, Journal of Materials Research, 19, 6,
(2004), pp. 1608-1612.
[18] S. Terashima and M. Tanaka, Thermal fatigue properties of Sn-1.2Ag-0.5Cu-xNi Flip
Chip interconnects, Materials Transactions, 45, 3, (2004), pp. 681-688.
Recrystallization
202
[19] S. K. Kang, P. A. Lauro, D.-Y. Shih, D. W. Henderson, and K. J. Puttlitz, Microstructure
and mechanical properties of lead-free solders and solder joints used in
microelectronic applications, IBM Journal of Research and Development, 49, 4/5,
(2005), pp. 607-620.
[20] A. U. Telang, T. R. Bieler, J. P. Lucas, K. N. Subramanian, L. P. Lehman, Y. Xing, and E.
J. Cotts, Grain-boundary character and grain growth in bulk tin and bulk lead-free
solder alloys, Journal of Electronic Materials, 33, 12, (2004), pp. 1412-1423.
[21] L. P. Lehman, S. N. Athavale, T. Z. Fullem, A. C. Giamis, R. K. Kinyanjui, M.
Lowenstein, K. Mather, R. Patel, D. Rae, J. Wang, Y. Xing, L. Zavalij, P. Borgesen,
and E. J. Cotts, Growth of Sn and intermetallic compounds in Sn-Ag-Cu solder,
Journal of Electronic Materials, 33, 12, (2004), pp. 1429-1439.
[22] Z. G. Chen, Y. W. Shi, Z. D. Xia, and Y. F. Yan, Study on the microstructure of a novel
lead-free solder alloy SnAgCu-RE and its soldered joints, Journal of Electronic
Materials, 31, 10, (2002), pp. 1122-1128.
[23] Maik Mller, Steffen Wiese, and Klaus-Jrgen Wolter, Influence of cooling rate and
composition on the solidification of SnAgCu solders, The Proceedings of the 1st
Electronics Systemintegration Technology Conference, Dresden, Germany, September 5
- 7, 2006, IEEE/EIA CPMT, (2006), pp. 1303-1311
[24] O. Fouassier, J.-M. Heintz, J. Chazelas, P.-M. Geffroy, and J.-F. Silvain, Microstructural
evolution and mechanical properties of SnAgCu alloys, Journal of Applied
Physics, 100, 043519 (2006), pp. 043519-1 - 043519-8.
[25] J. Gong, C. Liu, P. P. Conway, and V. V. Silberschmidt, Crystallographic structure and
mechanical behaviour of SnAgCu solder interconnects under a constant loading
rate, The Proceedings of the 57th Electronic Component and Technology Conference,
Reno, NV, May 29- June 1, 2007, IEEE/EIA CPMT, (2007), pp. 677-683
[26] P. Lauro, S. K. Kang, W. K. Choi, and D.-Y. Shih, Effects of mechanical deformation
and annealing on the microstructure and hardness of Pd-free solders, Journal of
Electronic Materials, 32, 12, (2003), pp. 1432-1440.
[27] M. Krause, M. Mueller, M. Petzold, S. Wiese, and K. J. Wolter, Scaling effects on grain
size and texture of lead free interconnects Investigations by electron backscatter
diffraction and nanointendation, The Proceedings of the 58th Electronic Component
and Technology Conference, Orlando, FL, May 27-30, 2008, IEEE/EIA CPMT, (2008),
pp. 75-81.
[28] T.-M. K. Korhonen, P. Turpeinen, L. P. Lehman, B. Bowman, G. H. Thiel, R. C. Parkes,
M. A. Korhonen, D. W. Henderson, and K. J. Puttlitz, Mechanical properties of
near-eutectic Sn-Ag-Cu alloy over a wide range of temperatures and strain rates,
Journal of Electronic Materials, 33, 12, (2004), pp. 1581-1588.
[29] W. Peng, K. Zeng, and J. Kivilahti, A literature review on potential lead-free solder
systems, Espoo, Helsinki University of Technology, Report Series HUTEPT1,
(2000), 53 p.
[30] S. Chada, R. A. Fournelle, W. Laub, and D. Shangguan, Copper substrate dissolution
in eutectic Sn-Ag solder and its effect on microstructure, Journal of Electronic
Materials, 29, 10, (2000), pp. 1214-1221.
[31] M. O. Alam, Y. C. Chan, and K. N. Tu, Effect of 0.5 wt% Cu addition in Sn3.5%Ag
solder on the dissolution rate of Cu metallization, Journal of Applied Physics, 94, 12,
(2003), pp. 7904-7909.
[32] W. G. Bader, Dissolution of Au, Ag, Pd, Pt, Cu and Ni in a Molten Sin-Lead Solder,
Welding Journal, 48, 12, (1969), pp. 551s-557s.
The Failure Mechanism of Recrystallization Assisted Cracking of Solder Interconnections
203
[33] W. G. Bader, Dissolution and formation on intermetallics in the soldering process, The
proceedings of the Conference on Physical Metallurgy and Metal Joining, St. Louis, MO,
Oct. 16-17. 1980, Warrendale, USA.
[34] W. C. Leslie, T. J. Michalak, and F. W. Aul, "The annealing of coldworked iron," in C.
W. Spencer and F. E. Werner, Iron and Its Dilute Solid Solutions. New York, 1963,
Interscience Puhlishers, pp. 103-119.
[35] R. W. Cahn, "Recovery and recrystallization," in R. W. Cahn, Physical Metallurgy,
Amsterdam, 1965, NorthHolland Publishing Company, pp. 925987.
[36] F. J. Humphreys and M. Hatherly, Recrystallization and Related Annealing Phenomena, 2
nd
ed., Oxford, 2004, Elsevier Ltd., 574 p.
[37] R.D. Doherty, D. A. Hughes, F. J. Humphreys, J. J. Jonas, D. Juul Jensen, M. E. Kassner,
W. E. King, T. R. McNelley, H. J. McQueen, and A. D. Rollett, Current issues in
recrystallization, Materials Science and Engineering A, 238, (1997), pp. 219 274.
[38] H. Yu and J. K. Kivilahti, Nucleation kinetics and solidification temperatures of
SnAgCu interconnections during reflow process, IEEE Transactions on
Components and Packaging Technologies, 29, 4, (2006), pp. 778 - 786.
[39] R. Darveaux, C. Reichman, and P. Agrawal, Solidification behavior of lead free and tin
lead solder bumps, The Proceedings of the 60th Electronic Component and Technology
Conference, Las Vegas, NV, June 1-4, 2010, IEEE/EIA CPMT, (2010), pp. 1442-1447.
[40] J. S. Kang, R. A. Gagliano, G. Ghosh, and M. E. Fine, Isothermal solidification of Cu/Sn
diffusion couples to form thin-solder joints, Journal of Electronic Materials, 31, 11,
(2002), pp. 1238-1243.
[41] J.-M. Song, J.-J. Lin, C.-F. Huang, and H.-Y. Chuang, Crystallization, morphology and
distribution of Ag
3
Sn in SnAgCu alloys and their influence on the vibration
fracture properties, Materials Science and Engineering A, 466, (2007), pp. 9-17
[42] D. W. Henderson, T. Gosselin, A. Sarkhel, S. K. Kang, W.-K. Choi, D.-Y. Shih, C.
Goldsmith, and K. J. Puttlitz, Ag
3
Sn plate formation in the solidification of near
ternary eutectic SnAgCu alloys, Journal of Materials Research, 17, (2002), pp. 2775-
2778.
[43] J. H. Perepezko, D. H. Rasmussen, I. E. Anderson, and C. R. Loper, Jr., Undercooling of
low-melting point metals and alloys, The Proceedings of the International Conference
on Solidification and Casting of Metals, Sheffield, England, July 1977, Sheffield
Metallurgical and Engineering Association / University of Sheffield / the Metals
Society, (1979), pp. 169-174.
[44] S. Wiese, E. Meusel, and K. J. Wolter, Microstructural dependence of constitutive
properties of eutectic SnAg and SnAgCu solders, The Proceedings of the 53
rd
Electronic Components and Technology Conference, 27 May-30 May, 2003, New
Orleans, LA, IEEE EIA/CPMT, (2003), pp. 197-206.
[45] S. K. Kang, W. K. Clioi, D.-Y. Shih, D. W. Henderson, T. Gossefin, A. Sarkliel, C.
Goldsmith, and K. J. Puttlitz, Formation of Ag
3
Sn plates in Sn-Ag-Cu alloys and
optimization of their alloy composition, The Proceedings of the 53
rd
Electronic
Components and Technology Conference, 27 May-30 May, 2003, New Orleans, LA,
IEEE EIA/CPMT, (2003), pp. 64-70.
[46] D. Swenson, The effects of suppressed beta tin nucleation on the microstructural
evolution of lead-free solder joints, Journal of Material Science: Materials in
Electronics, 18, 1-3, (2007), pp. 39-54.
[47] S. K. Kang, G. C. Moon; P. Lauro, and D.-Y. Shih, Critical Factors Affecting the
Undercooling of Pb-free, Flip-Chip Solder Bumps and In-situ Observation of
Recrystallization
204
Solidification Process, The Proceedings of the 57th Electronic Component and
Technology Conference, Reno, NV, May 29-June 1, 2007, IEEE/EIA CPMT, (2007), pp.
1597-1603.
[48] J. W. Elmer, E. D. Specht, and M. Kumar, Microstructure and in situ observations of
undercooling for nucleation of -Sn relevant to lead-free solder alloys, Journal of
Electronic Materials, 39, 3, (2010), pp. 273-282.
[49] M. A. Matin, W. P. Vellinga, and M. G. D. Geers, Thermomechanical fatigue damage
evolution in SAC solder joints, Materials Science and Engineering A, 445-446, (2007),
pp. 73-85.
[50] N. S. Brar and W. R. Tyson, Elastic and plastic anisotropy of white tin, Canadian
Journal of Physics, 50, 19, (1972), pp. 2257-2264.
[51] Metals Handbook, Volume 1 Properties and Selection of Metals, 8th ed., New York,
1961, American Society for Metals, 1300 p.
[52] K. N. Subramanian, Role of anisotropic behavior of Sn on thermomechanical fatique
and fracture of Sn-based solder joints under thermal excursions, Fatigue and
Fracture of Engineering Materials and Structures, 30, 5, (2007), pp. 420-431.
[53] M. A. Matin, E. W. C. Coenen, W. P. Vellinga, and M. G. D. Geers, Correlation between
thermal fatigue and thermal anisotropy in a Pb-free solder alloy, Scripta Materialia,
53, (2005), pp. 927-932.
[54] T. R. Bieler, H. Jiang, L. P. Lehman, T. Kirkpatrick, E. J. Cotts, and B. Nandagopal,
Influence of Sn grain size and orientation on the thermomechanical response and
reliability of Pb-free solder joints, IEEE Transactions on Components and Packaging
Technologies, 31, 2, (2008), pp. 370-380.
[55] T. R. Bieler, B. Zhou, L. Blair, A. Zamiri, P. Darbandi, F. Pourboghrat, T.-K. Lee, and K.-
C. Liu, The role of elastic and plastic anisotropy of Sn on microstructure and
damage evolution in lead-free solder joints, The Proceedings of the 2011 IEEE
International Reliability Physics Symposium, 10-14 April, 2011, Monterey, CA, IEEE,
(2011), pp. 5F.1.1-5F.1.9.
[56] S. Terashima, K. Takahama, M. Nozaki, and M. Tanaka, Recrystallization of Sn grains
due to thermal strain in Sn-1.2Ag-0.5Cu-0.05N solder, Materials Transactions, Japan
Institute of Metals, 45, 4, (2004), pp. 1383-1390.
[57] S. Dunford, S. Canumalla, and P. Viswanadham, Intermetallic morphology and
damage evolution under thermomechanical fatigue of lead (Pb)-free solder
interconnections, The Proceedings of the 54th Electronic Components and
Technology Conference, June 1-4, 2004, Las Vegas, NV, USA, IEEE/EIA/CPMT,
(2004), pp. 726-736.
[58] L. Lehman, S. Athavale, T. Fullem, A. Giamis, R. Kinyanjui, M. Lowenstein, K. Mather,
R. Patel, D. Rae, J. Wang, Y. Xing, L. Zavalij, P. Borgesen, and E. Cotts, Growth of
Sn and intermetallic compounds in Sn-Ag-Cu solder, Journal of Electronic Materials,
3, 12, (2004), pp. 1429-1439.
[59] P. Limaye, B. Vandevelde, D. Vandepitte, and B. Verlinden, Crack growth rate
measurement and analysis for WLCSP Sn-Ag-Cu solder joints, The proceedings of
the SMTA international annual conference, Chicago, IL, September 25-29, (2005),
pp. 371-377.
[60] L. Xu and J. H.L. Pang, Intermetallic growth studies on SAC/ENIG and SAC/CU-OSP
lead-free solder joints, Thermal and Thermomechanical Phenomena in Electronics
Systems, (2006), pp.1131-1136.
The Failure Mechanism of Recrystallization Assisted Cracking of Solder Interconnections
205
[61] A. U. Telang, T. R. Bieler, A. Zamirini, and F. Pourboghrat, Incremental
recrystallization/crain growth driven by elastic strain energy release in a
thermomechanically fatigued lead-free solder joint, Acta Materialia, 55, (2007), pp.
2265-2277.
[62] J. J. Sundelin, S. T. Nurmi, and T. K. Lepist, Recrystallization behavior of SnAgCu
solder joints, Materials Science and Engineering A, 474, (2008), pp. 201-207.
[63] J. Li, J. Karppinen, T. Laurila, and J. K. Kivilahti, Reliability of Lead-Free solder
interconnections in Thermal and Power cycling tests, IEEE Transactions on
Components and Packaging Technologies, 32, (2009), pp. 302-308.
[64] H. T. Chen, M. Mueller, T. T. Mattila, J. Li, X.W. Liu, K.-J. Wolter, and M. Paulasto-
Krckel, Localized Recrystallization and Cracking of Lead-Free Solder
Interconnections under Thermal Cycling, Journal of Materials Research, 25, 16,
(2011), pp. 2103-2116.
[65] T. T. Mattila, M. Mueller, M. Paulasto-Krckel, and K. J. Wolter Failure mechanism of
solder interconnections under thermal cycling conditions, The Proceedings of the 3rd
Electronic System-Integration and Technology Conference, Berlin, Germany, September
13-16, 2010, IEEE/EIA CPMT, (2010), pp. 1-8.
[66] B. Zhou, T. T. Bieler, T. K. Lee, and K.C. Liu, Crack development in a low-stress PBGA
package due to continuous recrystallization leading to formation of orientations
with [001] parallel to the interface, Journal of Electronic Materials, 39, (2010), pp.
2669 -2679.
[67] D. Hardwick, C. M. Sellars, and W. J. McG. Tegart, The occurrence of recrystallization
during high-temperature creep, Journal of the Institute of Metals, 90, (1961), pp. 21-22.
[68] D. McLean and M. H. Farmer, The relation during creep between grainboundary
sliding, sub-crystal size, and extension, Journal of Institute of Metals, 85, (1956), pp.
41-50.
[69] P. Gay and A. Kelly, X-ray studies of polycrystalline metals deformed by rolling. II.
Examination of the softer metals, tin, zinc, lead and cadmium, Acta
Crystallographica, 6, (1953), pp. 172-177.
[70] A. G. Guy, Elements of Physical Metallurgy, 2nd ed., London, 1960, Addison-Wesley
Publishing Company Inc., 528 p.
[71] S. Miettinen, Recrystallization of Lead-free Solder Joints under Mechanical Load, Masters
Thesis (in Finnish), Espoo, (2005), 84 p.
[72] T. M. Korhonen, L. Lehman, M. Korhonen, and D. Henderson, Isothermal fatigue
behavior of the near-eutectic Sn-Ag-Cu alloy between -25C and 125C, Journal of
Electronic Materials, 36, 2, (2007), pp. 173-178.
[73] T. Laurila, T. T. Mattila, V. Vuorinen, J. Karppinen, J. Li, M. Sippola, and J. K. Kivilahti,
Evolution of microstructure and failure mechanism of lead-free solder
interconnections in power cycling and thermal shock tests, Microelectronics
Reliability, 47, 7, (2007), pp. 1135-44.
[74] T. T. Mattila, H. Xu, O. Ratia, and M. Paulasto-Krckel, Effects of thermal cycling
parameters on lifetimes and failure mechanism of solder interconnections, The
Proceedings of the 60th Electronic Component and Technology Conference, Las Vegas,
NV, June 1-4, 2010, IEEE/EIA CPMT, (2010), pp. 581-590.
[75] H. Xu, T. T. Mattila, O. Ratia, and M. Paulasto-Krckel, Effects of thermal cycling
parameters on lifetimes and failure mechanism of solder interconnections, IEEE
Transactions on Manufacturing and Packaging Technologies a Special Issue, (in print).
Invited paper.
Recrystallization
206
[76] T. T. Mattila and J. K. Kivilahti, The role of recrystallization in the failure mechanism
of SnAgCu solder interconnections under thermomechanical loading, IEEE
Transactions on Components and Packaging Technologies, 33, 3, (2010), pp. 629-635.
[77] J. Li, T. T. Mattila, and J. K. Kivilahti, Multiscale simulation of recrystallization and
grain growth of Sn in lead-free solder interconnections, Journal of Electronic
Materials, 39, 1, (2010), pp. 77-84.
[78] J. Li, H. Xu, T. T. Mattila, J. K. Kivilahti, T. Laurila, and M. Paulasto-Krckel,
Simulation of dynamic recrystallization in solder interconnections during thermal
cycling, Computational Materials Science, 50, (2010), pp. 690-697.
9
Mathematical Modeling of Single Peak Dynamic
Recrystallization Flow Stress Curves in
Metallic Alloys
R. Ebrahimi and E. Shafiei
Department of Materials Science and Engineering, School of Engineering,
Shiraz University, Shiraz,
Iran
1. Introduction
Deformation of metals and alloys at temperatures above 0.5 T
m
is a complex process in
which mechanical working interacts with various metallurgical processes such as dynamic
restoration including recovery and recrystallization, and phase transformation for
polymorphous materials. The understanding of these processes, however, enables the
behavior of the metals and alloys. Recent developments have been described in several
review papers. These reviews were in agreement that metals and alloys having relatively
low values of stacking fault energy (SFE) could recrystallize dynamically, whereas those of
high SFE including bcc metals and alloys which behave in a manner similar to fcc metals of
high SFE recovered dynamically only during high temperature deformation. So that,
according to microstructural evolutions, material response can principally be divided into
two categories in hot deformation: dynamic recovery (DRV) type and dynamic
recrystallization (DRX) type.
The process of recrystallization of plastically deformed metals and alloys is of central
importance in the processing of metallic alloys for two main reasons. The first is to soften
and restore the ductility of material hardened by low temperature deformation that
occurring below about 50% of the absolute melting temperature which leads to lower forces.
The second is to control the microstructure and mechanical properties of the final product.
The analysis of metal forming process such as hot rolling, forging and extrusion has been
dependent on various parameters including constitutive relation which contributes to stress-
strain curves at high temperatures, shape of workpiece and product, shapes of tools, friction,
temperature, forming speed, etc. In such parameters, the constitutive equation is one of the
most important factors which have an effect on solution accuracy.
A number of research groups have attempted to develop constitutive equations of materials
and suggested their own formulations by putting the experimentally measured data into
one single equation. Misaka and Yoshimoto proposed a model which gives the flow stress of
carbon steels as a function of the strain, strain rate, temperature and carbon content. Shidas
model takes account of the flow stress behavior of the steels in austenite, ferrite and in the
Recrystallization
208
two-phase regions. Voce suggested an approximate equation of stressstain curve
considering the dynamic recrystallization. Finally, Ebrahimi et al. obtained a mathematical
model according to the phenomenological representation of the shape of the stress-strain
curves that consists an additional constant. Due to the importance of flow stress estimation
of metals and alloys at high temperatures, the required forces for the deformation processes,
dimensional accuracy of final products and simulation of processes; it is necessary to
provide a model in order to eliminate the limitations of previous models to some extent. The
contexts of this section could generally be classified into two main parts, at the beginning,
some previous mathematical models and basic concepts in relation with dominate processes
during hot deformation of metals and alloys, such as dynamic recovery (DRV) and dynamic
recrystallization (DRX) will be reviewed and finally, by concluding from existing
mathematical models related to prediction of single peak flow stress curves, a new model
will be introducd. With regards to the importance of macroscopic data from mechanical
tests as compared to microscopic ones from metallurgical investigation, due to its less time
and cost consuming nature, mathematical and macroscopic aspects of DRX process are
considered in this model.
2. Basic concepts
2.1 Dynamic recovery (DRV)
The basic mechanisms of dynamic recovery are dislocation climb, cross-slip and glide, which
result in the formation of low angle boundaries as also occurs during static recovery. However,
the applied stress provides an additional driving pressure for the movement of low angle
boundaries and those of opposite sign will be driven in opposite directions, and this stress-
assisted migration of dislocation boundaries may contribute significantly to the overall strain.
Such migration results in some annihilation of dislocations in opposing boundaries and Y-
junction boundary interactions and these enable the subgrains to remain approximately
equiaxed during the deformation. In-situ SEM deformation experiments have shown that
some reorientation of subgrains may also occur during hot deformation. The subgrains can
therefore be considered to be transient microstructural features.
The processes of work hardening and recovery lead to the continual formation and
dissolution of low angle boundaries and to a constant density of unbound or free
dislocations within the subgrains. After a strain of typically 0.5 to 1, the subgrain structure
often appears to achieve a steady state. The microstructural changes occurring during
dynamic recovery are summarized schematically in figure 1.
2.2 Dynamic recrystallization (DRX)
In metals which recovery processes are slow, such as those with a low or medium stacking
fault energy (copper, nickel and austenitic iron), dynamic recrystallization may take place
when a critical deformation condition is reached. A simplified description of the phenomenon
of dynamic recrystallization is as follows. As shown in figures 2, new grains originate at the
old grain boundaries, however, as the material continues to deform, the dislocation density of
the new grains increases, thus reducing the driving force for further growth, and the
recrystallizing grains eventually cease to grow. An additional factor which may limit the
growth of the new grains is the nucleation of further grains at the migrating boundaries.
Mathematical Modeling of Single Peak
Dynamic Recrystallization Flow Stress Curves in Metallic Alloys
209
Fig. 1. Summary of the microstructural changes which occur during dynamic recovery.
This type of dynamic recrystallization, which has clear nucleation and growth stages, can
be classified as a discontinuous process. There are other mechanisms which produce high
angle grain boundaries during high temperature deformation and which may be
considered to be types of dynamic recrystallization such as continuous dynamic
recrystallixation (CDRX).
The general characteristics of dynamic recrystallization are as follows (Humphreys et al.,
Elsevier):
- As shown in figure 3, the stress-strain curve for a material which undergoes dynamic
recrystallization generally exhibits a broad peak that is different to the plateau,
characteristic of a material which undergoes only dynamic recovery (fig.1). Under
conditions of low ZenerHollomon parameter, multiple peaks may be exhibited at low
strains, as seen in figure 3.
- A critical deformation is necessary in order to initiate dynamic recrystallization.
- The critical deformation decreases steadily with decreasing stress or ZenerHollomon
parameter (description in section 3.2), although at very low strain rates (creep) the
critical strain may increase again.
- The size of dynamically recrystallized grains increases monotonically with decreasing
stress. Grain growth does not occur and the grain size remains constant during the
deformation.
- The flow stress and grain size are almost independent of the initial grain size, although
the kinetics of dynamic recrystallization are accelerated in specimens with smaller
initial grain sizes.
- Dynamic recrystallization is usually initiated at pre-existing grain boundaries although
for very low strain rates and large initial grain sizes, intragranular nucleation becomes
more important.
Recrystallization
210
Fig. 2. Evolution of microstructure during hot deformation of a material showing DRX.
Fig. 3. The effect of Zener Hollomon parameter on the stress- strain curve.
Mathematical Modeling of Single Peak
Dynamic Recrystallization Flow Stress Curves in Metallic Alloys
211
2.3 Test methods
There are many different types of test methods, such as hot torsion, compression, tension,
cam plastometer and drop hammer for developing the constitutive equation of the materials
behavior. Among them, hot torsion and compression tests have widely been used. Hot
torsion test has been used by many researchers to formulate constitutive equation of
materials subject to a large strain because it has a forte in simulating the multi-pass
deformation, in comparison with axisymmetric compression test. In the case of compression
test, high friction at the interface of material and stroke head results in barreling during test.
Thus, the compression test has a limitation in generating flow stress curve when the
material undergoes large strain. Also Kim et al. have shown that the flow stress obtained
from compression test is higher than that from torsion test, although the general shapes of
the measured curves are similar. The differences between the stresses were approximately in
the range of 1020%.
These differences might be attributed to the following reasons: compression test has the
glowing frictional forces at the ram-specimen interface as the test progress, while, there is
no frictional effect in torsion test. In compression test, it is difficult to achieve constant
strain rate and isothermal condition during test, whereas to control them in torsion test is
accurate.
2.4 Flow curves
For metals with high stacking fault energy which experience DRV, the flow stress curves
increase with strain in the initial deformation and reach constant in consequence of attaining
the balance between work hardening and DRV (saturation stress,
s
). For metals with DRX,
initially the flow stress increases with strain which is being dominated by work hardening,
and as DRX takes place upon critical strain (
c
), the flow stress begins to decrease after
reaching certain peak value. When the equilibrium is reached between softening due to DRX
and work hardening, the curves drop to a steady state region (
ss
). As shown in figure 3, the
stress strain curves of a dynamically recrystallizing material may be characterized by a
single peak or by several oscillations. Luton and Sellars have explained this in terms of the
kinetics of dynamic recrystallization. At low stresses, the material recrystallizes completely
before a second cycle of recrystallization begins, and this process is then repeated. The flow
stress, which depends on the dislocation density, therefore oscillates with strain. At high
stresses, subsequent cycles of recrystallization begin before the previous ones are finished,
the material is therefore always in a partly recrystallized state after the first peak, and the
stress strain curve is smoothed out, resulting in a single broad peak. Fig. 4 shows a
schematic representing of dynamic recovery and a single peak dynamic recrystallization.
2.5 Strain hardening rate versus stress
The change in the slope (=dd) of the stress-strain curve with stress can be a good
indication of the microstructural changes taking place in material. All of the - curves for a
particular alloy originate from a common intercept
0
at =0. The - curves have three
segments, two of them are linear. The first linear segment decreases with stress for initial
strain to the point where subgrains begin to form with a lower rate of increase in DRV. The
curve gradually changes to the lower slope of the second linear segment up to the point
Recrystallization
212
where the critical stress
c
is attained for initiating the dynamic recrystallization. The curve
then drops off rather sharply to =0 at peak stress. Extrapolation of the second linear
segment of the - curve intercepts the axis at the saturation stress. This would be the
shape of the - curve if dynamic recrystallization were absent with only dynamic recovery
as the restorative mechanism operation, Fig.5.
Fig. 4. Schematic representing of DRV and DRX.
Fig. 5. Changes in the slope of the stress-strain curve with stress.
Mathematical Modeling of Single Peak
Dynamic Recrystallization Flow Stress Curves in Metallic Alloys
213
3. A review on presented models
Considerable researches have been carried out to model flow stress curves at high
temperatures based on both empirical and mathematical models. Empirical models which
are based on trial and error mainly obtained by repeating the mechanical tests at different
conditions, and among these models, Shida and Misaka models could be mentioned. The
most important limitation of these types of models is that they are valid for specific
conditions and cannot be generalized, whereas in mathematical models, in order to estimate
the flow curve, the material response are modeled first and then the empirical data is used
to verify the prediction. Among these models, Cingara, Kim et.al and Ebrahimi et.al could
be mentioned. According to the presented models, some limitations could be considered in
this category, too. For instance, the presented model by McQueen and Cingara for DRX is
only valid up to the peak stress (
p
) and does not consider the softening behavior due to
DRX, although a complement model has been developed to predict the softening behavior
due to DRX.
Mathematical modeling of flow stress curves can be divided into three main topics as
follows:
- Mathematical models which characterize the initiation of DRX.
- Mathematical models which predict characteristic points of flow stress curves.
- Mathematical models which predict flow stress curves.
It is important to mention that, due to less time and cost consuming nature of mathematical
and macroscopic models, only this aspect of DRX process is considered in this chapter.
3.1 Initiation of dynamic recrystallization
The critical strain for initiation of DRX could be determined by metallography. However,
this technique requires extensive sampling before and after the critical strain. Furthermore,
phase changes during cooling from hot working temperature alter the deformed structure,
which in turn render difficulties for metallographic analysis. Also this technique requires a
large number of specimens deformed to different strains. On the other hand, the critical
strain thus obtained is not precise.
Several attempts have been made to predict the initiation of DRX. For example, Ryan and
McQueen observed that the presence of a stress peak at a constant strain rate flow curve
leads to an inflection in plots of strain hardening rate, , versus stress, . Moreover, the
points of inflection in - plots where the experimental curves separate from the
extrapolated lower linear segments give critical conditions for initiation of DRX. Later,
Poliak and Jonas have shown that this inflection point corresponds to the appearance of
an additional thermodynamic degree of freedom in the system due to the initiation of
DRX.
3.1.1 Determination of critical stress
The simple method of Najafizadeh and Jonas was used for determination of the critical
stress for initiation of DRX. The inflection point is detected by fitting a third order
polynomial to the - curves up to the peak point as follows:
Recrystallization
214
3 2
= + + + u o o o A B C D
(1)
where A, B, C, and D are constants for a given set of deformation conditions. The second
derivative of this equation with respect to can be expressed as:
2
2
6 2 = +
u
o
o
d
A B
d
(2)
At critical stress for initiation of DRX, the second derivative becomes zero. Therefore,
6 2 0
3
+ = = o o
c c
B
A B
A
(3)
An example of - curves and its corresponding third order polynomial are shown in Fig. 6.
Therefore, this method is used to determine the value of critical stress at different
deformation conditions. Using the flow curves, the values of critical strain are determined
(Najafizadeh et al. 2006).
According to Fig. 7, the normalized critical stress and strain for 17-4 PH stainless steel could
be presented as:
0.89 =
o
o
c
p
(4)
0.467 =
c
c
c
p
(5)
Fig. 6. The - curve of 17-4 PH Stainless steel and its corresponding third order polynomial.
Mathematical Modeling of Single Peak
Dynamic Recrystallization Flow Stress Curves in Metallic Alloys
215
Fig. 7. Critical stress and strain versus (a) peak stress and (b) peak strain.
3.1.2 Determination of critical strain, (Poliak and Jonas)
Poliak and Jonas have shown that the inflection in plots of ln- and ln- curves can also be
used for determination of initiation of DRX. The procedure of Section 3.1.1 is used to determine
the values of critical strain at different deformation conditions (Mirzadeh et al., 2010).
According to figure 8, the normalized critical strain could be presented for 17-4 PH stainless
steel:
0.47 =
c
c
c
p
(6)
Fig. 8. The ln- curve for 17-4 PH stainless steel and its corresponding third order
polynomial.
Recrystallization
216
3.1.3 Determination of critical strain (Ebrahimi and Solhjoo)
The flow stress up to the peak stress was modeled using the Cingara and McQueen equation
as shown below:
[( / )exp(1 / )] = o o c c c c
C
p p p
(7)
Where constant C is an additional parameter to make the results acceptable and is obtained
from logarithmic form of Eq.7. The derivative of the true stress with respect to true strain
yields the work hardening rate, . Therefore, the formula using Eq.7 may be expressed as:
(1 / 1 / ) = u o c c
p
C
(8)
In order to determine the critical strain, the second derivative of with respect to must be
zero. By solving this equation, the critical strain as a function of peak strain will obtain as
follows:
1 (1 )
=
c
c
c
p
C C
C
(9)
The results showed a good agreement with the experimentally measured ones for Nb-
microallyed steel (Ebrahimi et al., 2007).
3.2 Kinetic equations (prediction of single points)
The influence of temperature and strain rate on peak stress was analyzed by the following
equations which were originally developed for creep but have found applicability in the
high strain rates in hot working:
''
.
'
exp
| |
= =
|
\ .
o c
n HW
p
Q
A Z
RT
(10)
.
''
exp( ) exp
| |
= =
|
\ .
|o c
HW
p
Q
A Z
RT
(11)
.
''' sinh( exp
| |
(
= =
|
\ .
oo c
n
HW
p
Q
A Z
RT
(12)
where A
,A
,A''',n,n
,Q
HW
,,
and R are constants and Z is the Zener Hollomon. Q
HW
is the
activation energy related to hot working and R is the universal gas constant. A power law
plot of log versus log Z gives linear segments only at low stress, indicating the limited
applicability of Eq. (10). However in a plot of against log Z linearity is lost at low stress,
showing that Eq. (11) is an inadequate fit for the entire range of stresses. Eq. (12) is a more
general form of Eq. (10) and Eq. (11) reducing to Eq. (10) at lower stresses (<0.8) and Eq.
(11) at higher stresses (>1.2). The validation of these equations to predict characteristic
points of flow curves were approved by many authors (Cingara et al., 1992).
Mathematical Modeling of Single Peak
Dynamic Recrystallization Flow Stress Curves in Metallic Alloys
217
3.3 Experimental models
3.3.1 Misakas model
Misaka and Yoshimoto have utilized the following double-power constitutive equation to
determine flow stress associated with the processing of steels:
| | | |
( )
| | | |
2
.
2
2851 2968 1120
9.8exp 0126 1.75 0.594
273
m
n
Misaka
C C
C C
T
o c c
| |
+
|
= + +
| +
\ .
(13)
Application range of this formula is as follows; carbon content: 1.2%, temperature: 750
1200
o
C, reduction (natural strain): 50% and strain rate: 30200 s
-1
. Misakas equation was
updated by including effects of solution-strengthening and dynamic recrystallization.
The updated Misakas constitutive equation is:
( )( )
,
1
Misaka Updated Misaka DRX ss DRX
f X k X o o o = + (14)
| | | | | | | | | | | |
0.8 0.3
0.835 0.51 0.098 0.128 0.144 0.175 0.01 = + + + + + + f Nb Mn Cr Mo V Ni (15)
(
Misaka
)updated indicates the flow stress of steels containing multiple alloying additions.
ss
is steady state stress, K =1.14 is a parameter that converts flow stress to mean flow stress,
and X
DRX
is volume fraction of dynamic recrystallization. It might be useful for practical
purpose but its mathematical base is weak. The factors for Mn, Nb, V and Ni are linear
although the terms for Cr and Mo are nonlinear. Devadas et al. compared the predicted flow
stress data for a low alloy steel with measured data from a cam-plastometer. They showed
that Misakas model overestimated the flow stress (Kim et al., 2003).
3.3.2 Shidas equation
Shidas equation is based on experimental data obtained from compression type of high
strain rate testing machines. Shida then expressed the flow stress of steels, , as a function of
the equivalent carbon content (C), the strain (), the strain rate (
.
c ) and temperature (T) as
followings:
( ) ( )
.
,
| |
=
|
\ .
o o c c
d w r
C T f f (16)
5 0.01
0.28exp
0.05
| |
=
|
+
\ .
o
d
T C
(17)
| |
273
1000
(
+
=
o
T C
T k (18)
( ) 1.3 0.3
0.2 0.2
| | | |
=
| |
\ . \ .
c c
c
n
w
f (19)
Recrystallization
218
0.41 0.07 = n C
(20)
.
.
10
| |
| |
|
=
|
|
\ . |
\ .
c
c
m
r
f
(21)
( ) ( ) 0.019 0.126 0.076 0.05 = + + m C T C
(22)
where f
w
(strain) and f
r
(strain rate) are functions dependent upon strain and strain rate,
respectively. The formulation of Eq. (16) is based on assumption that flow stress increases
with the strain rate and strain increased. The range of validity of the formula is quite broad.
This formula is applicable in the range of carbon content: 0.071.2%, temperature:
7001200C, strain: 0.7 and strain rate: 100 s
1
.
3.3.3 Modified Voces equation
In contrast to Misakas and Shidas equations, Voces equation can describe the flow stress
over the wide range of strains and strain rates. The equation can express the dynamic
softening portion of the flow stress curve by using Avrami equation:
( ) 1 exp
+
= (
o o c
m
WH DRV p
c
(23)
The coefficient, C, and work hardening exponent, m, are dependent on the deformation
conditions. The parameters, C and m are normally taken as being a constant, however, it is a
function of the deformation conditions (strain rate, temperature). Thus, the C and m are
considered to be a function of dimensionless parameter, Z/A.
Also Kim et al. developed an equation by modifying Voces constitutive equations
accounting for the dynamic recrystallization as well as the dynamic softening. During
thermomechnical processing, the important metallurgical phenomena are work hardening
(WH), dynamic recovery (DRV) and dynamic recrystallization (DRX). Thus, the flow stress
curve can be bisected with the WH + DRV region and the DRX region. For the evaluation of
the WH and DRV region, Eq. 23 was used. Also For the region of DRX, the drop of flow
stress was expressed as the following equation:
( )
1
p
p
DRX
Drop p ss
X X
X
c
c
o o o
(
( =
(
for >
p
(24)
Therefore, the flow stress can be expressed in subtraction form as follows:
+
= o o o
WH DRV Drop
(25)
where
ss
is the steady state stress achieved at larger strains and X
p
the volume fraction of
DRX at peak strain. X
DRX
is the volume fraction of DRX at any strain.
Mathematical Modeling of Single Peak
Dynamic Recrystallization Flow Stress Curves in Metallic Alloys
219
Fig. 9 shows the measured and predicted flow stress curves in large strain range of AISI
4140 steel when the three different types of constitutive equations are used for prediction.
The flow stress curves calculated by using Misakas equation agree with the measured ones
with some extent of error in comparison to the flow stress curve obtained from Shidas
equation. Although, the stressstrain curves predicted by using the modified Voces
equation are in a good agreement with experimentally measured ones, it seems Misakas
equation does not reflect recrystallization behavior properly (Kim et al., 2003).
Fig. 9. Comparison of measured and predicted constitutive relations for AISI 4140 steel.
3.4 Mathematical models
3.4.1 Solhjoos model
Solhjoo was considered linear estimation of - curve up to the peak stress as follows:
4 1
c
= +
c
o
c
S A
(26)
where S
4
is the slope of the line and A
1
is a constant. Due to the first assumption which
considers a linear segment up to the peak, this model would not be able to show the critical
strain. Using the maximum point of the - curve (=
p
,=0), value of constant A
1
is
obtained to be S
4
p
. Solution of the differential Eq. (26) with respect to , using boundary
condition =
p
at =
p
is:
( )
( )
4
1 exp
( | |
( |
=
( |
\ .
c c
o o c
c
p
p
p
S
(27)
Recrystallization
220
where
p
is the peak stress and
p
is the peak strain.
Determination of S
4
can be simply done using a linear plot of ln
| |
|
|
\ .
c o o
o c c
p p
p p
vs. ; S
4
would
be the slope of the line. As S
4
is very sensitive to strain rate, it is better to plot ln
| |
|
|
\ .
c o o
o c c
p p
p p
vs. for each sets of strain rates, and then the average value of the slopes determines S
4
.
Assuming S
4
as a constant shows a rough estimation of stress strain curve. Since S
4
is more
sensitive to strain rate than temperature, this parameter can be considered as a power law in
form of:
.
4
= c
E
S C
(28)
where E is strain rate and C and are constants. Using a plot of ln(-S4) vs.
.
c , constants C
and E can be determined. It should be considered that another limit of this model is its
disability of prediction of flow stress at very low strains (less than 0.05) that the work
hardening rate has very high values (Solhjoo, 2009).
3.4.2 Avramis analysis
The DRX may be considered as a solid-state transformation and its kinetics can be modeled
by the JohnsonMehlAvramiKolmogorov (JMAK) equation as follows:
( )
1 exp
n
DRX
X kt = (29)
where X
DRX
and t are the recrystallized volume fraction and DRX time, respectively. The
effect of dynamic recovery (DRV) on flow softening was not considered. Therefore, this
model is preliminary intended for materials with relatively low stacking fault energy.
Moreover, for modeling the flow curves after the peak point of stressstrain curve, the
initiation of DRX was intentionally considered at peak point. This assumption simplifies the
Avrami analysis with acceptable level of accuracy. Therefore, Eq. (30) gives the magnitude
of flow stress at each fractional softening:
( )
p p ss DRX
X o o o o =
(30)
3.4.3 Ebrahimis model
This model is based on a phenomenological representation of the shape of the flow stress
curves and the traditional theories for constitutive equations which incorporate the power
law. Ebrahimi et al. considered the variations of the slope of flow stress curves as follows:
( )
1
1
| |
| =
|
\ .
o c
o o
c c
ss
p
d
C
d
(31)
Mathematical Modeling of Single Peak
Dynamic Recrystallization Flow Stress Curves in Metallic Alloys
221
Where
p
is the strain at the peak stress and
ss
is the steady state stress. The term (1-/
p
)
estimates variation of the stress- strain curve for >
ss
. Solution of the differential Eq.(31)
with respect to using boundary condition
p
at
p
results in:
( )
2
1
exp
2 2
( | |
( | = +
|
(
\ .
c
c
o o o o c
c
p
ss p ss
p
C (32)
If
k
and
k
k
p
where k<1 and
k
>
ss
, then coefficient C
1
can be estimated from Eq. (33)
as:
( )
1
2
2
ln
2 1
p ss
k ss
p
C
k k
o o
o o
c
| |
=
|
|
+
\ .
(33)
Where
k
is the stress calculated from Cingara equation. The equation represented by this
model required the values of stress and strain at the peak and stress at the steady state zone.
These parameters can be calculated by kinetic equations (Ebrahimi et al., 2006).
Fig. 10 shows the calculated flow curves by Cingara, Avrami and Ebrahimi et al. equations.
At low Z, the predictions of Ebrahimi et al. equation are relatively accurate (Fig. 10a), but at
high Z, this equation overestimates the flow softening of DRX (Fig. 10c). In other words, this
equation seems to be suitable for ideal DRX behavior. The Ebrahimi et al. equation is based
on a phenomenological representation of the shape of the stressstrain curves and the
traditional theories for constitutive equations which incorporate the power law. Conversely,
the Avrami equation is amenable to all deformation conditions as shown in Fig. 10. In fact,
this equation is an adaptive one with two adjustable constants. As a result, it could be better
fitted to hot flow curves. Therefore, the Avrami equation can be used for prediction of flow
curves as shown in Fig. 10 (Mirzadeh et al., 2010).
3.4.4 Shafiei and Ebrahimis constitutive equation
Using the extrapolation of DRV flow stress curves and kinetic equation for DRX, Shafiei and
Ebrahimi proposed the following equation for modeling single peak DRX flow curves for
c ss
c c c s <
( ) ( )
'
exp( ) ( )
s s c c s ss DRX
C X o o o o c c o o =
(34)
Where C is a constant with metallurgical sense. According to the geometrical relations
shown in Fig.5, the value of C can be formulized as Eq.s (35) and (36) .
c
,
p
,
ss
,
s
,
c
,
X
DRX
are critical stress, peak stress, steady state stress, saturation stress , critical strain and
volume fraction of DRX, respectively.
'
1
tan
C
=
(35)
2 ( )
p p
c c p
Arctg
o c
t
c c c
| |
| = +
|
\ .
(36)
Recrystallization
222
As shown in Fig.11, the stress-strain curves predicted by using presented model are in a
good agreement with experimentally measured ones for Ti-IF steel. In order to evaluate the
accuracy of the model, the mean error was calculated. The mean error of flow stress is
calculated at strains of 0.19-2 for every measurement under all deformation conditions. For
all stress-strain curves, the mean errors are between -2.9 and 2.5. The results indicate that
the proposed model give a good estimate of the flow stress curves. Therefore, it can be
deduced that the approach to obtain a constitutive equation applicable to large strain ranges
was fruitful and this presented model might have a potential to be used where more precise
calculation of stress decrement due to dynamic recrystallization is important. Moreover, this
analysis has been done for the stress-strain curves under hot working condition for Ti-IF
steel, but it is not dependent on the type of material and can be extended for any condition
that a single peak dynamic recrystallization occurs.
Fig. 10. Comparison between calculated and measured flow stress curves of 17-4 PH
stainless steel.
Mathematical Modeling of Single Peak
Dynamic Recrystallization Flow Stress Curves in Metallic Alloys
223
Fig. 11. Comparison of measured and predicted stress-strain curves of Ti-IF steel.
4. Summary
In this chapter, a review on recent model of single peak flow stress curves was presented. At
first, the basic concepts on hot deformation and dynamic restoration, including affecting
factors on dominated processes such as DRX and DRV and related microstructure
evolutions were discussed. Then, an introduction on experimental models which are more
capable in the field of industrials investigations such as Misaka, Shida and Voce constitutive
equations followed by the details of mathematical models such as MC Queen, Ebrahimi,
Solhjoo, Avrami and Shafiei- Ebrahimi constitutive equations were presented. In this case,
the accuracy of these models as well as some limitations was evaluated in order to obtain
the optimum working conditions.
5. References
Cingara, A. & McQueen, H. J. (1992). New method for determining sinh constitutive
constants for high temperature deformation of 300 austenitic steel. Mater. Proc.
And Tech, 36, 17-30.
Ebrahimi, R. & Solhjoo, S. (2007). Characteristic points of stress-strain curve at high
temperature. ISSI, 2, 24-27.
Ebrahimi, R., Zahiri, S.H. & Najafizadeh, A. (2006). Mathematical modeling of the stress-
strain curves of TI-IF steel at high temperature. Mater. Proc. And Tech., 171, 301-
305.
Ebrahimi, R. (2003). Hot working of Ti-IF steel. Ph.D thesis. Isfahan University of
Technology.
Recrystallization
224
He, X. Yu, Zh. & Lai, X. (2008). A method to predict flow stress considering dynamic
recrystallization during hot deformation. Comp. Mater. Sci., 44, 760-764.
Humphreys, F.J. & Hatherly, M. (2004). Recrystallization and related annealing phenomena
(Second edition), (Elsevier, UK).
Imbert, C. A. C & McQueen, H. J. (2001). Peak strength, strain hardening and dynamic
restoration of A2 and M2 tool steels in hot deformation. Mater. Sci. and Eng. A, 313,
88-103.
Imbert, C. A. C & McQueen, H. J. (2001). Dynamic recrystallization of A2 and M2 tool steels.
Mater. Sci. and Eng. A, 313, 104-116.
Jonas, J. J., Quelennec, Jiang, L. & Martin, E. (2009). The avrami kinetics of dynamic
recrystallization. Acta Mater. 137, 1748-2756.
Kim, S. I., Lee, Y. & Byon, S. M. (2003). Study on constitutive relation of AISI 4140 steel
subject to large strain at elevated temperature. Mater. Proc. And Tech.,140, 84-89.
Liu, J., Cai, Zh. & Li, C. (2008). Modelling of flow stress characterizing dynamic
recrystallization for magnesium alloys. Comp. Mater. Sci., 41,375-382.
Mirzadeh, H. & Najafizadeh, A. (2010). Extrapolation of flow curves at hot working
conditions. Mater. Sci. and Eng. A, 527,1856-1860.
Mirzadeh, H. & Najafizadeh, A. (2010). Prediction of the critical conditions for initiation of
dynamic recrystallization. Mater. And Des., 31, 1174-1179.
Najafizadeh, A. & Jonas, J. J. (2006). Predicting the critical stress for initiation of dynamic
recrystallization. ISIJ Int., 46, 1679-1684.
Poliak, E. I. & Jonas, J. J. (2002). Initiation of dynamic recrystallization in constant strain rate
hot deformation. ISIJ Int., 43, 684-691.
Shaban, M. & Eghbali, B. (2010). Determination of critical conditions for dynamic
recrystallization of a microalloyed steel. Mater. Sci. and Eng. A, 527, 4320-4325.
Shafiei, E. & Ebrahimi, R. (2011). Mathematical modeling of single peak flow stress curves.
Comp. Mater. Sci., submitted paper.
Solhjoo, S. (2009). Analysis of flow stress up to the peak at hot deformation. Mater. And
Des.,30, 3036-3040.
Ueki, M., Horie, S. & Nakamura, T. (1987). Factors affecting dynamic recrystallization of
metals and alloys. Mater. Sci. and Tech.,3, 329.
Verlinden, B., Driver, J., Samijdar, I. & Doherty, R. D. (2007). Thermomechanical processing
of metallic materials. (Elsevier, Uk).
Zahiri, S. H., Davies, C. H. J. & Hodgson, P. D. (2005). A mechanical approach to quantify
dynamic recrystallization in polycrystalline metals. Scripta Mater., 52, 299-304.
Zeng, Zh., Jonsson, S., Rovan, H. J. & Zhang, Y. (2009). Modelling the flow stress for single
peak dynamic recrystallization. Mater. And Des.,30-1939-1943.
10
Phase Transformations and Recrystallization
Processes During Synthesis, Processing and
Service of TiAl Alloys
Fritz Appel
Institute for Materials Research, Helmholtz-Zentrum Geesthacht, Geesthacht,
Germany
1. Introduction
Titanium aluminides alloys based on the intermetallic phases o
2
(Ti
3
Al) and (TiAl) are one
of the few classes of emerging materials that have the potential for innovative applications
in advanced energy conversion systems whenever low density, good high-temperature
strength and resistance against ignition and corrosion are of major concern [1]. The
outstanding thermo-physical properties of the individual phases mainly result from the
highly ordered nature and directional bonding of the compounds. However, two-phase
o
2
(Ti
3
Al)+(TiAl) alloys exhibit a much better mechanical performance than their monolithic
constituents (TiAl) and o
2
(Ti
3
Al), provided that the phase distribution and grain size are
suitably controlled. The synergistic effects of the two phases are undoubtedly associated
with the many influences that the microstructure has on deformation and fracture processes.
Constitution and microstructure are the result of phase transformations, ordering reactions
and recrystallization processes, which occur during synthesis, processing and service. Many
aspects of these mechanisms are intimately linked to defect configurations at the atomic
level; thus standard techniques of metallography were often inadequate to provide the
necessary information. This lack of information is addressed in the present article in that
observations on recrystallization and phase transformations by high-resolution electron
microscopy are presented. Particular emphasis will be paid on
i. the origin of microstructures
ii. heterogeneities in the deformed state and recovery behaviour
iii. atomic structure of crystalline and crystalline/amorphous interfaces
iv. misfit accommodation and coherency stresses.
2. Constitution and microstructure
2.1 Constitution
TiAl alloys of technical significance have the general composition (in at. %, as are all
compositions in this paper)
Ti-(42-49)Al+X, (1)
Recrystallization
226
with X designating alloying elements, such as Cr, Nb, W, V, Ta, Si, B, and C [2]. When
referred to the binary phase diagram (Fig. 1), the equilibrium phases for Al contents
between 46 and 49 % are: the disordered solution phases hexagonal (h.c.p.) o(Ti), body
centred cubic (b.c.c.) |(Ti), and the ordered intermetallic compounds (TiAl) with L1
0
structure, and o
2
(Ti
3
Al) with D0
19
structure [3]. Based on this constitution, different
microstructures have been designed. The phase transformations involved in the
microstructural evolution will be demonstrated for two examples, the classical lamellar
structure of o
2
(Ti
3
Al)+(TiAl) alloys and a novel crystallographically modulated
morphology occurring in multiphase alloys.
Fig. 1. Central portion of the Ti-Al-phase diagram in the region of technical interest [3].
2.2 Lamellar microstructure
The technologically most relevant o
2
(Ti
3
Al)+(TiAl) alloys contain a significant volume
fraction of lamellar grains [1, 2]. The morphology of these grains represents a multilayer
system made of two phases. It is well documented in the literature [4] that such a system
could exhibit extraordinary mechanical properties when the layer thickness is small enough.
The so-called lamellar microstructure results from the precipitation of lamellae in either a
disordered o or a congruently ordered o
2
matrix, following one of the transformation paths [3]:
oo
2
o
2
+ or oo
2
+. (2)
o
2
and o
2
have the same crystal structure but different composition. The exact pathway is
still a matter of debate and could depend on alloy composition and thermal treatment. The
orientation relationships between the o
2
and platelets are [5]
Phase Transformations and Recrystallization
Processes During Synthesis, Processing and Service of TiAl Alloys
227
{111} || (0001)o2 and [1 1 0] || <11 2 0>o2. (3)
The length of the lamellae is determined by the size of the parent o/o
2
grain. The phase is
formed as an ordered domain structure, as sketched in Fig. 2. This gives rise to six variants
of the above orientation relationships. Thus, there are four types of lamellar interfaces: the
o
2
/ interface and three distinct / interfaces that are typified by rotations of 60, 120 and
180 between adjacent lamellae. A lamellar grain consists of a set of lamellae, which are
subdivided into domains and interspersed by o
2
lamellae. The volume fraction of the two
phases is controlled by the composition on the basis of the phase diagram and the
processing conditions of the alloy. However, it should be noted that the decomposition of o
phase into lamellar (o
2
+) is sluggish and often cannot be established within the constraints
of processing routes; thus the volume fraction of phase is less than equilibrium [6]. It might
be expected that the Ti concentration in the phase increases when the alloy becomes richer
in Ti until the maximum solubility of Ti in the phase is reached. This non-equilibrium
phase composition may provide significant driving forces for structural changes, as will be
outlined in the subsequent sections.
Fig. 2. Schematic drawing of a lamellar colony comprised of and o
2
platelets.
Among the various lamellar interfaces only the twin boundary is fully coherent, as the
adjacent lattices are symmetrically oriented. At all the other interfaces the matching is
imperfect, i.e., these interfaces are semicoherent. The mismatch arises from the differences in
the crystal structure and lattice parameters and amounts to 1 to 2 %, depending on alloy
composition and processing conditions. Different modes of mismatch accommodation have
Recrystallization
228
been discussed for lamellar TiAl alloys, which in broad terms correspond to the early
models of misfitting interfaces [7]. Up to a certain point, the misfit strain could be solely
taken up by elastic distortion, i.e., the lamellae are uniformly strained to bring the atomic
spacings into registry. This homogeneous strain accommodation leads to coherent interfaces
but introduces lattice distortions that are known as coherency strains. Hazzledine [4] has
shown that the elastic misfit accommodation in lamellar (o
2
+) alloys is only possible if the
lamellae are very thin. The predicted critical thicknesses are d
c
8 nm for the mismatched /
interfaces and d
c
0.8 to 3.9 nm for the o
2
/ interfaces, depending on the volume content of
o
2
phase. Figure 3 shows a small Ti
3
Al platelet embedded in phase with a thickness of 4.5
nm, which is just above the coherency limit predicted by Hazzledine [4]. While
homogeneous strain accommodation is still recognizable, part of the misfit is already taken
up by interfacial defects.
Fig. 3. High-resolution electron micrograph showing a small Ti
3
Al platelet embedded in
phase. Interfacial steps, dislocations and homogeneous elastic straining accommodate the
misfit between the particle and the matrix. In the compressed image below, the elastic
straining at the tip of the particle is readily visible by the distortion of the ( 111 ) planes. A
dislocation compensating the misfit between the ( 111 )
and {2 2 01}
o2
planes is arrowed. Ti-
46.5Al-4(Cr, Ta, Mo, B), sheet material.
The coherency strains raise the total energy of the system. Thus, for a sufficiently large
misfit, or lamellar spacing, it becomes energetically more favourable to replace the coherent
interface by a semicoherent interface, a situation that is referred to as a loss of coherency. As
Phase Transformations and Recrystallization
Processes During Synthesis, Processing and Service of TiAl Alloys
229
described in the early model of Frank and van der Merve [8] misfit dislocations partially
take up the misfit, i.e., the atoms at the interface adjust their positions to give regions of
good and bad registry. In other words, the misfit is concentrated at the dislocations.
Mismatch strains at interfaces can be also relieved by the formation of ledges or steps at the
interface [9]. The introduction of mono-atomic steps significantly improve the atomic
matching, thus preventing the disregistry from becoming large anywhere. Structural ledges
may replace misfit dislocations as a way of retaining low-index terraces between the
respective defects. The general view is that planar boundaries are favoured for large misfits
and small Burgers vectors of the misfit dislocations, whereas stepped boundaries are
favoured for small misfits and large values of the Burgers vector. Steps may range in scale
from atomic to multi-atomic dimensions depending on energetic or kinetic factors.
However, it is very often the case that an interfacial defect exhibits both dislocation- and
step-like character, thus, comprising a more general defect that has been defined as a
disconnection [10,11]. Because of its step character, disconnection motion along an interface
transports material from one phase to the other, the extent of which is essentially
determined by the step height. At the same time, the dislocation content of the
disconnection leads to deformation. In this sense, disconnection motion couples interface
migration with deformation. The different extents of symmetry breaking at the interfaces
lead to a broad variety of disconnections [10, 11]. This is reflected in different step heights
and dislocation contents, which eventually determine the function of disconnections in
phase transformations. Disconnection models have been developed for a variety of
diffusional and diffusionless phase transformations in crystalline solids; and an extensive
body of literature has evolved. For more details the reader is referred to a review of Howe et
al. [11]. Several authors [12-16] observed interfacial steps at o
2
/ interfaces with heights that
were always a multiple of {111}
and t(o
2
)=[000 1 ]
o2
[17]. b is the Burgers vector of the dislocation component
(parallel to the (111) interface) and t() and t(o
2
) are vectors describing the ledge risers [11] of
the disconnection. This type of disconnection has no Burgers vector component
perpendicular to the interface.
The complexity of misfit accommodation at o
2
/ interfaces is illustrated in Fig. 4. The
micrograph shows a (TiAl) lamella terminated within the o
2
phase of a two-phase alloy.
The interface marked in the micrograph borders the crystal region in which the exact ABC
stacking of the L1
0
structure is fulfilled. Outside of this exactly stacked region of phase,
there is a two to three atomic plane thick layer in which neither the ABC stacking of the
phase nor the ABAB stacking of the o
2
phase is correctly fulfilled. This becomes particularly
evident at the tip of the lamellae and indicates a significant homogeneous straining of the
lattice. This strain seems to locally relax by the formation of dislocations, as can be seen in
the compressed form of the image. The other salient feature is the misfit accommodation by
steps. A Burgers circuit constructed around the tip of the lamella results in a projected
Burgers vector of b
p
=1/6[ 112 ]. This indicates that the small misfit between the (111)
and
(0002)
o2
planes is elastically taken up. The observed misfit accommodation is pertinent to
the issue of how differential material flux during the o
2
transformation is accomplished.
Recrystallization
230
There is ample evidence [18] of enhanced self-diffusion along dislocation cores. Likewise
interfacial ledges are envisaged as regions where deviation from the ideal structure is
localized and which may provide paths of easy diffusion. This gives rise to the speculation
that it is mainly the tip of a newly formed lamella where the atomic composition between
the two phases is adjusted during transformation.
Fig. 4. A high-resolution micrograph of a (TiAl) lamella terminated within o
2
phase imaged
down the <10 1 >
projection, which can be used as a reference. The interface between the lath and the
phase (designated as /T) consists mainly of flat terraces, which are parallel to the (111)
plane.
Steps of different heights often delineate the terraces. The modulated laths are comprised of an
orthorhombic constituent, which is interspersed by slabs of |/B2 and a little o
2
phase (Fig. 6).
Selected area diffraction of the orthorhombic constituent is consistent with the B19 phase,
which can be described as the orthorhombic phase (oP4) or as a hexagonal superstructure of
D0
19
(hP8). In the Ti-Al system, the B19 structure has already been observed by Abe et al. [23]
and Ducher et al. [24]. The B19 structure is structurally closely related to the orthorhombic
phase (oC16, Cmcm) with the ideal stoichiometry Ti
2
AlNb, which among the intermetallic
compounds is remarkable for its relatively good room temperature ductility [25].
Recrystallization
232
In Figs. 5b and 6a the B19 structure is imaged in the [010]
B19
projection. As deduced from the
high-resolution images, the orientation relationships between the constituents involved in a
modulated lath with the adjacent are [22]
Fig. 5. TEM analysis of the modulated microstructure. (a) A modulated lath imaged by
diffraction contrast. (b) High-resolution TEM micrograph of a modulated lath adjacent to a
lamella.
Phase Transformations and Recrystallization
Processes During Synthesis, Processing and Service of TiAl Alloys
233
Fig. 6. High-resolution TEM evidence of the modulated microstructure. (a) and (b) Fourier-
filtered images of the areas boxed in Fig. 5.
(100)
B19
|| {110}
|/B2
|| (0001)
o2
|| {111}
[010]
B19
|| <111>
|/B2
|| <11 2 0>
o2
|| <1 1 0>
; [001]
B19
|| <11 2 >
.
(5)
In the diffraction pattern the presence of the periodic distortion is manifested by the
existence of weak satellite reflections adjacent to the main reflections. The distance of the
satellites from the main reflections is the reciprocal of the modulation wavelength, and the
direction joining the satellites with their main reflections is parallel to the direction of the
modulation vector. These observed features are reminiscent of a modulated structure, which
in recent years have attracted considerable interest; for a review see [26]. A crystal structure
is said to be modulated if it exhibits periodicities other than the Bravais lattice periodicities.
These additional periodicities arise from one or more distortions, which increase to a
maximum value and then decrease to the initial value. The modulation may involve atomic
coordinates, occupancy factors or displacement parameters. The strain of the discontinuities
is often relieved by a continuous and periodic variation of the physical properties of the
Recrystallization
234
product. First principle calculations of Nguyen-Manh and Pettifor [27] and Yoo and Fu [28]
have shown that the |/B2 phase existing in TiAl alloys containing supersaturations of
transition metals (Zr, V, Nb) is unstable under tetragonal distortion; a shear instability that
was attributed to the anomalous (negative) tetragonal shear modulus. Specifically, B2 may
transform by homogeneous shear to several low temperature orthorhombic phases, which
can exist metastably. The energetically favourable transformations are [27]
B2(Pm3m) B19(Pmma) B33(Cmcm). (6)
At the atomic level, the B2 phase may transform to B19 by a shuffle displacement of
neighboring (011)
B2
planes in opposite [01 1 ] directions, as illustrated in Fig. 7. A
subsequent displacement of neighboring (011)
B2
planes in the [100] direction generates the
B33 structure [27]. In the system investigated here the predominant orthorhombic phase
seems to be B19. It is tempting to speculate that the modulation of the laths is triggered by a
periodic variation of the composition of the parent B2 phase, as occurs during spinodal
decomposition. Clearly the mechanism requires further investigation.
Fig. 7. Formation of the B19 structure from the parent B2 phase illustrated by perspective
views of hard-sphere models. [100] projection of the B2 phase; arrow heads mark shuffle
displacements of neighbouring (011)
B2
planes in opposite [01 1 ] directions to form B19.
The modulated laths can apparently further transform into the phase, as demonstrated in
Fig. 8. This process often starts at grain boundaries and proceeds through the formation of
high ledges via distinct atomic shuffle displacements. As this transformation was frequently
observed in deformed samples [22], it might be speculated that the process is stress induced
and provides some kind of transformation toughening.
Phase Transformations and Recrystallization
Processes During Synthesis, Processing and Service of TiAl Alloys
235
Fig. 8. Stress-induced transformation of a modulated lamella (T) into phase. (a) Generation
of a lamella at a grain boundary. (b) Higher magnification of the area marked by the arrow
in (a). Note the ledges at the interfaces.
3. Hot working
Titanium aluminide alloys are relatively brittle materials; attaining chemical homogeneity and
refinement of the microstructure are therefore the most important prerequisites for
engineering applications. To this end, large effort has been expended to establish wrought
processing of TiAl alloys [29, 30]. Hot working of these alloys is generally impeded by a
significant plastic anisotropy [31], low diffusivity [32] and the susceptibility to hot cracking
[29]. While these aspects are well documented in the TiAl literature, there are many open
questions about the elementary processes that determine dynamic recovery, recrystallization
and phase transformations. These aspects are addressed in the present section.
Recrystallization
236
3.1 The deformed state
A commonly held concept of hot working is that dynamic recovery and recrystallization are
triggered by heterogeneities in the deformed state. Thus, the deformation mechanisms
occurring in the majority phases (TiAl) and o
2
(Ti
3
Al) will briefly be discussed, for details see
[1, 31]. Deformation of (TiAl) is mainly provided by ordinary dislocations with the Burgers
vector b=1/2<110] and mechanical twinning along 1/6< 112 ]{111}. Since twinning shear is
unidirectional, the operating twinning systems vary with the sense of the load and the loading
direction. There are crystal orientations for which twinning is forbidden. To a lesser extent
deformation is provided by the motion of superdislocations with the Burgers vectors b=<101]
and b=1/2< 112 ]. The superdislocations exhibit an asymmetric non-planar core spreading;
this results in a high glide resistance, which is sensitive to the direction of motion, for a review
see [33]. Thus, the superdislocations do not move as readily as ordinary dislocations and
mechanical twins. Taken together, this glide geometry gives rise to a significant plastic
anisotropy of the phase. Ti
3
Al alloys have several potential slip systems, which in principle
may provide sufficient shear components for the deformation of polycrystalline material.
However, there is a strong predominance for prismatic slip; plastic shear with c components of
the hexagonal cell is practically impossible [34]. This plastic anisotropy appears to be even
more enhanced at intermediate temperatures because the pyramidal slip systems exhibit an
anomalous increase of the critical resolved shear stress with temperature. Thus, the brittleness
of polycrystalline Ti
3
Al alloys can be attributed to the lack of independent slip systems that can
operate at comparable stresses; hence the von Mises criterion is not satisfied. Due to this
situation deformation of (TiAl)+o
2
(Ti
3
Al) alloys is mainly carried by the phase.
There are various processes that might lead to heterogeneities in the deformed state. As a
specific example for TiAl alloys, deformation heterogeneities resulting from mechanical
twinning will be outlined in more detail. At the beginning of deformation the slip path of
the twins is essentially identical with the domain size or lamellar spacing. As soon as
multiple twinning with non-parallel shear vectors is activated, extensive intersections
among twin bands occur. The intersection of a moving deformation twin with a barrier twin
is expected to be difficult because the incorporation of an incident twinning system into the
barrier twin may no longer constitute a crystallographically allowed twinning system.
Several authors [35-39] have analyzed the mechanism and have proposed crystallographic
relations by which an incident twin could intersect a barrier twin. The high-resolution
electron micrograph shown in Fig. 9 demonstrates a so-called type-I twin intersection, which
is favoured if the < 110 ] intersection line is at 0 to 55 from the sample axis [35]. The
structure is imaged along the common < 110 ] direction of the two twins; this can be
recognized by the different contrast of the (002) planes, which are alternately occupied by Ti
and Al atoms. On its upper side the vertical twin is thicker than on its lower side, thus this
twin was considered as incident twin T
i
; T
b
is the barrier twin. The intersection leads to a
significant deflection of the two twins, which, however, is more pronounced for the barrier
twin. The difficulty in forming a twin intersection is manifested by the ledged interface
between the incident twin and the matrix (arrows 1 and 3) and various dislocations that
emerge at the intersection zone (arrows 2). Figure 10 shows the details of the intersection
zone. The intersection zone remains in the L1
0
structure and seems to be relatively free of
defects; however highly defective regions border it. The (002)
Tb
planes of the central zone
Phase Transformations and Recrystallization
Processes During Synthesis, Processing and Service of TiAl Alloys
237
Fig. 9. Type-I intersection of two deformation twins observed after room temperature
compression of a Ti-48.5Al-0.37C alloy. The structure is imaged along the common < 110 ]
intersection line. The vertical twin T
i
is considered to be the incident twin because its upper
side is thicker than its lower side; T
b
is the barrier twin.
Recrystallization
238
Fig. 10. Structural details of the intersection zone: translation of the twinning shear along the
( 111 )
Tb
planes by 1/2< 101 ] superpartials that become evident by extra (002)
Tb
planes. Two
of the dislocations and their extra half planes are indicated by dislocation symbols. Note the
anticlockwise rotation of the intersection zone with respect to the barrier twin by 10, which
is indicated by the traces of the respective ( 111 )
Tb
planes. The compressed image below
shows these features in more detail. Arrowheads mark the dislocation walls on either side of
the intersection zone.
are not continuous with those of the barrier twin, but appear displaced along the ( 111 )
Tb
planes of the barrier twin. This displacement is consistent with glide of 1/2< 101 ] or
1/2< 011 ] superpartials on ( 111 )
Tb
planes. These dislocations become manifest by extra
(002)
Tb
planes; a few of these dislocations and the orientation of their extra planes are
indicated by dislocation symbols. It is tempting to speculate that these dislocations were
Phase Transformations and Recrystallization
Processes During Synthesis, Processing and Service of TiAl Alloys
239
generated under the high stress concentration acting at the corners of the intersection
zone. It is worth adding that a shear accommodation by twinning along the ( 111 )
Tb
planes is not possible because this would require anti-twinning operations. The close
distance of these dislocations explains why the lattice of the intersection zone is not
congruent with that of the barrier twin, but rotated by about 10 against the barrier twin.
The image below is the micrograph compressed along the (002) planes and shows these
details more clearly. The observation largely reflects the strong rotation field that was
generated by the incident twin. The various dislocation reactions that could be involved
in the intersection process probably give rise to dislocation emission; a few of these
dislocations are marked with arrows 1 to 3. Twin intersections undoubtedly leave
significant internal stresses and dense defect arrangements bordering the intersection
zone. Under hot-working conditions these heterogeneities in the deformed state can be
the prevalent sites for recrystallization. At elevated temperatures rearrangement of the
dislocation walls surrounding the misoriented zone may occur by climb so that the
misorientation with respect to the surrounding matrix increases. The intersection zone is
transformed into a new grain of low internal energy, growing into deformed material
from which it is finally separated by a high-angle boundary. This process is certainly
driven by the release of stored energy. Thus, it might be expected that the structural
heterogeneities produced by twin intersections act as precursor for recrystallization.
Such processes are certainly beneficial for the conversion of the microstructure under
hot-working conditions.
3.2 Dynamic recovery
There is good consensus that recovery and recrystallization are competing processes as
both are driven by the stored energy of the deformed state. The extent of recovery is
generally expected to depend on the stacking fault energy, which, in turn, determines
dislocation dissociation. The ordinary dislocations, which mainly carry the deformation in
(TiAl), have a compact core because dissociation would involve a high-energy complex
stacking fault (CSF), which destroys the chemical environment of first neighbours in the
fault plane. This compact core structure makes cross slip and climb of the ordinary
dislocations relatively easy and is a good precondition for static and dynamic recovery.
The most convincing evidence for climb of ordinary dislocations has also been obtained
from in situ heating experiments performed inside the TEM [31]. The samples used were
pre-deformed at room temperature to a strain of c=3%; this introduced sufficient
dislocations for observation and certainly a small supersaturation of intrinsic point
defects. Fig. 11 demonstrates the change of the dislocation fine structure occurring upon
heating by a sequence of micrographs. The vacancies produced during room-temperature
deformation condense onto screw dislocations, causing them to climb into helices. TEM
observations performed after high-temperature deformation have also revealed a
remarkable instability of twin structures [31]. Due to these factors, the release of strain
energy by recovery is probably relatively easy and can account for the large reduction of
the flow stress that has been observed after annealing at moderately high temperatures.
Recovery lowers the driving force for recrystallization; thus, a significant amount of prior
recovery may retard the recrystallization kinetics. More details about deformation and
recovery are provided in [40].
Recrystallization
240
Fig. 11. Climb of ordinary dislocations during in situ heating inside the TEM at an
acceleration voltage of 120 kV. Note the formation of a helical dislocation and the growth of
prismatic dislocation loops. Cast Ti-48Al-2Cr, pre-deformation at room temperature in
compression to c=3%.
Phase Transformations and Recrystallization
Processes During Synthesis, Processing and Service of TiAl Alloys
241
3.3 Dynamic recrystallization
The kinetics of dynamic recrystallization of TiAl alloys depends on several factors; these
involve alloy composition, grain size of the starting microstructure, and hot working
parameters [41]. The conversion of the coarse-grained lamellar ingot structure is probably
the most difficult step in wrought processing. Thus, dynamic recrystallization occurring at
this stage will be described as a specific example for TiAl alloys.
There is undisputed evidence that slip transfer through lamellar boundaries is difficult [42].
Plastic strain resulting from a cooperative operation of several deformation modes can be
localized between the lamellar boundaries. These facts, combined with the flat plate
geometry of the lamellae (Fig. 2), cause a marked plastic anisotropy of lamellar material,
which affects the recrystallization behaviour [41]. The recrystallization kinetics is relatively
fast if the lamellae orientation is parallel to the deformation axis. Deformation in this
orientation apparently involves an element of instability, which is manifested by kinking of
the lamellae (Fig. 12) and is reminiscent of buckling of load-carrying structures. In terms of a
laminate model [43], the lamellar morphology may be considered as an ensemble of TiAl
and Ti
3
Al plates. When perfectly aligned with the compression axis, these columns are
highly stable under compression, as long as the axial load is below a critical value. Above
this critical load the equilibrium becomes unstable and the slightest disturbance will cause
the structure to buckle. In the lamellar structure an upsetting moment might develop by
lateral impinging of the lamellae by dislocation pile-ups or deformation twins. Furthermore,
the elastic response of the o
2
and phase upon loading is significantly different. Thus, the
tendency to instable buckling is expected to increase, if there is an inhomogeneous
distribution of o
2
and lamellae. The process probably starts with local bending of the
lamellae. From the curvature and thickness of the lamellae local strains can be deduced,
which are often larger than 10 % and lead to the formation dense dislocation structures and
of sub-boundaries (Figs. 13 and 14). Subsequent rotation and coalescence of these sub-grains
occurs apparently in such a way that kinking of the lamellae is accomplished. Due to
kinking, the lamellae are reoriented with respect to the deformation axis, which may
support shearing along the lamellar interfaces. In the regions of highest local bending
spheroidization and dissolution of the o
2
phase occur; this suggests that both the non-
equilibrium constitution and the local stress provide the driving pressure for the observed
phase transformation and recrystallization. All these aspects are manifested in Fig. 15, which
shows the initial state of grain nucleation in a kinked o
2
lamella.
Micromechanical modelling [43, 44] has shown that polycrystalline lamellar material
deforms very inhomogeneously. Localized stress concentrations have been recognized,
which developed within the polycrystal upon straining due to the variations of grain size,
shape and orientation. This led the authors to believe that constraints imposed by
neighbouring grains are a dominant factor in determining the flow behaviour. Even under
compression, very high tensile hydrostatic stresses are generated at the triple points of
colony boundaries. There is a strong tendency to develop shear bands, kink bands, lattice
rotations, and internal buckling of lamellae. The shear bands consist of extremely fine grains
and may traverse the whole work piece, often resulting in gross failure. From a mechanical
point of view, buckling failures do not depend on the yield strength of the material but only
on the dimensions of the structure and the elastic properties of the material. In lamellar
alloys of given lamellar spacing, the tendency to buckling and shear band formation
Recrystallization
242
increases with the axial length of plates. Thus, coarse-grained lamellar alloys seem to be
prone to deformation instabilities and shear band formation, which turns out to be one of
the prime problems in wrought processing of TiAl alloys.
Fig. 12. Kinked lamellae in Ti-47Al-5Nb-0.2B-0.2C. The lamellar ingot material was subject at
1270 C to a compression stress of o=(25.412.7) MPa fluctuating with 30 Hz. C indicates the
orientation of the compression axis, which is parallel to the image plane.
Phase Transformations and Recrystallization
Processes During Synthesis, Processing and Service of TiAl Alloys
243
Fig. 13. Sub-boundaries formed at kinked lamellae. Experimental details as for Fig. 12.
Recrystallization
244
Fig. 14. Atomic structure of a sub-boundary formed at kinked lamella; experimental details
as for Fig. 12. (a) Mixed ordinary dislocations situated in the sub-boundary (marked with
arrow 1 in Fig. 13) and (b) on of the dislocations shown in higher magnification; the
compressed image below shows the extra plane of the dislocation more clearly.
Phase Transformations and Recrystallization
Processes During Synthesis, Processing and Service of TiAl Alloys
245
Fig. 15. Initial state of grain nucleation at kinked o
2
lamellae. The image below shows one of
the grains in higher magnification. Note the shear processes occurring along the lamellae.
Ti-45Al-8Nb-0.2C, sheet material; viewing direction in the transverse direction of the rolling
plane.
Recrystallization
246
4. Service-induced phase transformation and recrystallization
4.1 Creep
At temperatures above 650 C damage of TiAl components may occur due to continuous
creep. The relevant mechanisms are numerous and synergistic, depending on the operation
conditions; for a review see [45]. Among the various microstructures that can be established
in polycrystalline (o
2
+) alloys, fully lamellar alloys are most creep resistant. However,
numerous investigations have demonstrated that the lamellar microstructure degrades upon
creep. This structural instability is a serious problem for long-time service of lamellar alloys.
In the following, a few examples of TEM analysis will be presented, which demonstrate the
complexity of the processes involved in the degradation of the lamellar morphology. Most
of these studies have been performed on samples that had been subject to long-term tensile
creep at 700 C under relatively low stresses of 80 to 140 MPa [46].
When compared with the microstructure of undeformed material, the lamellar interfaces in
crept samples were highly imperfect. Figure 16a demonstrates the formation of a high
interfacial step in a 60 pseudo twin boundary. The interfacial steps had often grown into
broad zones, which extended over about 200 nm perpendicular to the interface; Fig. 16b
demonstrates an intermediate stage of this growth process. Multiple-height ledges are
commonly observed after phase transformation and growth, and several mechanisms have
been proposed to explain the phenomenon [47]. Analogous to these models it is speculated
that the large ledges observed in the crept TiAl alloy arise from one-plane steps, which
moved under diffusional control along the interfaces and were piled up at misfit
dislocations. A misfit dislocation with a Burgers vector component out of the interfacial
plane is arrowed in Fig. 16a. Once a sharp pile up is formed, the configuration may
rearrange into a tilt configuration with a long-range stress field. This would cause further
perfect or Shockley partial dislocations to be incorporated into the ledge and would also
explain that in all cases the ledges were associated with misfit dislocations. The detailed
atomic structure of the macro ledges is not clear. As can be seen in Figs. 16, there is a
variation of the contrast in the ledges with a periodicity of three (111) planes. This is
reminiscent of the 9R structure, which is a phase that probably has a slightly higher energy
than the L1
0
ground state. The formation of the 9R structure is a well-known phenomenon
in many f.c.c. metals that exhibit twinning. Singh and Howe [48] have recognized the 9R
structure in heavily deformed TiAl. It should be noted, however, that similar three-plane
structures have been observed in a massively transformed, but undeformed Ti-48.7Al [49].
In this work the contrast phenomena have been interpreted as arising from overlapping
twin related variants. Nevertheless, the macro-ledges are a characteristic feature in the
microstructure of crept samples and represent at least a highly faulted L1
0
structure; the
question is only whether there is a periodicity in the fault arrangement. When the macro-
ledges grow further, it might be energetically favourable to reconstruct the L1
0
structure and
to nucleate a new grain. Figure 17 probably demonstrates an early stage of such a process.
The recrystallized grains usually have a certain orientation relationship with respect to the
parent lamellae; Fig. 18 indicates that the (001) planes of the recrystallized grain are
parallel to the ( 111 ) planes in the parent lamella
1
. There is a significant mismatch for this
orientation relationship, which is manifested by a high density of ledges and dislocations at
the (001)||( 111 ) interface. Recrystallization of ordered structures has been investigated in
Phase Transformations and Recrystallization
Processes During Synthesis, Processing and Service of TiAl Alloys
247
several studies [50, 51]. There is a drastic reduction in grain boundary mobility, when
compared with disordered metals. Recovery of ordered alloys is also complicated by the fact
that the ordered state has to be restored. In this respect it is interesting to note that the small
grain shown in Fig. 18 is completely ordered giving the impression that the ordering is
immediately established after grain nucleation or that nucleation occurred in the ordered
state. This might be a consequence of the fine scale of the lamellar microstructure and the
heterogeneous grain nucleation at the interfacial ledges. There are certainly crystallographic
constraints exerted by the parent lamellae adjacent to the ledges, which may control
nucleation and growth. Clearly, the process needs further investigation.
There is a significant body of evidence in the TiAl literature indicating that dissolution of o
2
lamellae occurs during creep [45, 46]. The phase transformation is probably driven by a non-
equilibrium constitution. High-temperature creep is expected to promote phase
transformation towards equilibrium constitution; thus, dissolution of o
2
and formation of
occurs [52]. The high-resolution micrograph shown in Fig. 19 supports this reasoning; there
is clear evidence that the density of steps at the o
2
/ interfaces is significantly higher than
that at the / interfaces, meaning that the o
2
lamella dissolves, whereas the lamellae are
relatively stable. The processes eventually end with the formation of new grains (Fig. 20)
and a more or less complete conversion of the lamellar morphology into a fine spheroidized
microstructure. The o
2
phase transformation is often associated with local deformation,
as suggested by Fig. 21. The micrograph shows two o
2
terminations that are connected by an
interface, thus the o
2
lamella is partially dissolved. Two twins were emitted at one of the
terminations and a dislocation with a Burgers vector out of the interface is present (Fig. 22).
The o
2
terminations have extremely small principal radii of curvature. The elimination of
such structural features reduces the surface energy and provides a driving force towards
further coarsening. The interface connecting the two o
2
terminations exhibits a fault
translation that corresponds to an intrinsic stacking fault. This is indicated by the stacking
sequence ABC B CAB. The nature of the interface between two phases is determined by their
structural relationships. There is a strong tendency for planes and directions with the
highest atomic densities to align across the interface. As suggested by Chalmers and Gleiter
[53], a better atomic fit at a boundary could result if atoms were moved away from
coincident sites by a rigid-body displacement of one grain relative to the other by a constant
displacement vector. Atomic modelling performed in this context (for a review see [33]) has
shown that the energy of lamellar interfaces could be minimized by a rigid-body translation
along the vectors f
APB
=1/2<10 1 ]
, f
SISF
=1/6[11 2 ]
and f
CSF
=1/6[ 2 11]
. These translations
correspond to the formation of an antiphase boundary (APB), a superlattice intrinsic
stacking fault (SISF) and a complex stacking fault (CSF), respectively, at the interface. In the
present case the translation vector is of type f
SISF
=1/6[11 2 ]
m
s ubgrai ns (d) grai ns (D)
Fig. 7. The variation of the subgrain/grain size in dependence of changes in torsion
frequency at constans: o =6, v = 0,015 mm/s, Ah=7 mm
grain size
no more than 1 m
0.19
0.37
0.32
0.12
0
0.1
0.2
0.3
0.4
0.5
0.2 0.8 1.6 1.8
Torsion frequency, Hz
A
r
e
a
f
r
a
c
t
i
o
n
Fig. 8. The variation in the area fraction of ultrafine grain in dependence of changes in
torsion frequency at constans: o =6, v = 0,015 mm/s, Ah=7 mm
The Effect of Strain Path on the Microstructure and
Mechanical Properties in Cu Processed by COT Method
277
0.23
0.4
0.42
0.3
0
0.1
0.2
0.3
0.4
0.5
0.2 0.8 1.6 1.8
Torsion frequency, Hz
F
r
a
c
t
i
o
n
5o 5o - 15o 15o
Fig. 9. The variation in the misorientation angle in dependence of changes in torsion
frequency at constans: o =6, v = 0,015 mm/s, Ah=7 mm
When the torsional frequency is 0,8 Hz and 1,6 Hz, the average diameters of the Cu grains
have about 400 nm. A distinct increase of the average diameter of the grain is seen when the
torsional frequency is increased from 1,6 Hz to 1,8 Hz. The grains have a size of about 600
nm. Roughly, it can be assumed that the subgrains are about 2 times smaller than grains. The
area fraction of the grains to the magnitude of 1 m, also appears to be beneficial for the
torsional frequencies 0,8 Hz and 1,6 Hz (especially for 1,6 Hz) (Fig.8.). The Cu grains which
have the ultrametric size occupy about 40% of the analyzed surfaces. The application of the
high torsional frequencies during the deformation process allows for obtaining about 40% of
the high- angle boundaries fractions (Fig.9). It means that more than a half of the deformed
structure is dominated by the subgrains.
Fig. 10. Microstructure of Cu after COT deformation at: f=0,2 Hz, o =6, v=0,015 mm/s and
Ah=7 mm
Recrystallization
278
Fig. 11. Microstructure of Cu after COT deformation at: 11. f=0,8 Hz, o =6, v=0,015 mm/s
and Ah= 7 mm
After the COT deformation with the torsional frequency f=0,2 Hz, the dislocation cell
structure and the DDWs dislocation walls with a high density of dislocation are observed
(Fig.10). The effects of arranging the dislocation structure are seen between the particular
DDWs (Fig.10). In general, deformation carried out when the torsional frequencies are small
makes the fraction of the narrow-angle boundaries (up to 5) reaching almost 50% (Fig.9.)
and the area fraction of the ultrafine-grains does not exceed 20% (Fig.8.).
It was observed that when the torsional frequency increased, the grain division into smaller
volumes happened more often (Fig.11). The result of this is the generation of the greater
amount of the LABs dislocation boundaries. A great amount of the tested areas of thin foils,
shows that after the deformation with frequencies f= 0,8 Hz and f=1,6 Hz in particular, the
dislocation polygonal walls and the grains with the HAB boundaries are created in the
structure (Fig.12). The vacancy clusters visible in Cu, are the effects of the dislocation
annihilation (Fig.13). The results of measuring the dislocation density (Fig.14) are also the
proof of the above mentioned.
Fig. 12. Microstructure of Cu after COT deformation at: f=1,6 Hz, o =6, v=0,015 mm/s and
Ah= 7 mm
The Effect of Strain Path on the Microstructure and
Mechanical Properties in Cu Processed by COT Method
279
Fig. 13. Microstructure of Cu after COT deformation at: f=1,6 Hz, o =6, v=0,015 mm/s and
Ah=7 mm. The vacancy clusters are arrows marked
Fig. 14. Microstructure of Cu after COT deformation at: f=1,8 Hz, o =6, v=0,015 mm/s and
Ah=7 mm
Fig. 15. The variation of the dislocation density in COTes samples with deferent value of
torsion frequency and constant: o =6, v=0,015 mm/s, Ah=7 mm
Recrystallization
280
Together with the increase of the torsional frequency, the decrease of dislocation density is
observed (Fig.15). When the frequency is 1,8 Hz the recovery begins to dominate the
structure what has a negative influence on the material that is being refined (Fig.14). The
transformation of dislocation tangles into polygonal boundaries and the dislocation
annihilation, have no impact on raising the amount of the wide-angle boundaries fractions
as well as on increasing the area fraction of the ultrafine-grains.
4.2 The compression velocity
In the COT process the compression velocity is the parameter which when increased, has an
influence on highering the value of the compression force and on lowering the values of
equivalent deformations. For example, the compression velocity 0,015 mm/s and 0,04 mm/s
while the rest of the parameters is constant: 1,6 Hz, Ah=7, o=6; the effective deformation c
f
is 120 and 45. Applying even higher compression velocities i.e. v= 0,1 mm/s and v= 0,6
mm/s causes a significant decrease of effective deformation to the corresponding values
c
f
=12 and c
f
=2.
Fig. 16. EBSD maps of Cu after COT processing: a) v=0,04 mm/s; b) v=0,1 mm/s; c) v=0,6
mm/s; at constant parameters: o =6, v=0,015 mm/s i Ah=7 mm
The deformation realized when the compression velocity is increasing from 0,015 mm/s to
0,04 mm/s while the rest of the parameters is constant: 1,6 Hz, Ah=7, o=6; has an positive
influence on the grain refining as well as on the increase of the homogeneity of the structure
(compare Fig.6c and Fig.16a). The structures in the prevailing part of the areas are equiaxed,
the grains bigger than 1 m are not produced, the amount of high-angle boundaries is high.
Whereas, the further increase of this parameter delays the reduction of the grain size and the
creation of high-angle boundaries (Fig.16 b,c). During the deformation with the velocity of
0,1mm/s the effects of the structure deformation are clearly visible (Fig.16b), but when the
a)
The Effect of Strain Path on the Microstructure and
Mechanical Properties in Cu Processed by COT Method
281
compression has the velocity of 0,6 mm/s only the fragments of the near-angle boundaries
attest to the insignificant deformation of the material (Fig.16c). A significant reduction of the
deformation realized by the increase of the compression velocity, does not lead to the grain
refining.
When it comes to the Cu refined-grain, the juxtaposition of the structural changes for two
velocities: 0,015 mm/s and 0,04 mm/s as well as of different torsional frequencies is quite
interesting (Fig.17). That is why the further structural analysis involved comparing some
selected ways of deformation.
Comparing the EBSD images obtained for Cu after the deformation with the compression
velocity of v= 0,015 mm/s and v= 0,04 mm/s, the constant torsional frequency 1,6 Hz and
the rest of the parameters also being constant (Fig.6c and Fig.16a), it is seen that the
deformation happening at higher compression velocity is an effective way of grain-refining.
It is also proven by the results of the quantity tests shown in the Figs.18-20. The average
diameter of the Cu grain after the increase of the compression velocity, decreases from 400
nm to 300 nm (Fig.18). The average diameter of the subgrains also decreases from 230 nm to
180 nm (Fig.18). The area fraction of the grains up to 1 m constitute 60% which is
advantageous, when the torsional frequency is 1,6Hz and the compression velocity is
v= 0,04 mm/s (Fig.19).
Fig. 17. EBSD maps of Cu after COT processing: a) f=0,8Hz; b) f=1,8Hz; at constant
parameters o =6, v=0,04 mm/s and Ah=7 mm
Recrystallization
282
When the compression velocity is higher, the high-angle boundaries are easily created and
they constitute about 50% of the analyzed Cu areas (Fig.20).
0.26
0.18
0.21
0.43
0.3
0.41
0
0.1
0.2
0.3
0.4
0.5
0.8 1.6 1.8
Torsion frequency, Hz
A
v
e
r
a
g
e
d
i
a
m
e
t
e
r
,
m
s ubgra i ns (d) gra i ns (D)
Fig. 18. The variation of the subgrain/grain size in dependence of changes in torsion
frequency at constans o =6, v = 0,04 mm/s, Ah=7 mm
grain size no more than 1 m
0.33
0,60
0.46
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.8 1.6 1.8
Torsion frequency, Hz
A
r
e
a
f
r
a
c
t
i
o
n
Fig. 19. The variation in the area fraction of ultrafine grain in dependence of changes in
torsion frequency at constans: o =6, v = 0,015 mm/s, Ah=7 mm
0.29
0.52
0.41
0
0.1
0.2
0.3
0.4
0.5
0.6
0.8 1.6 1.8
Torsion frequency, Hz
F
r
a
c
t
i
o
n
5o 5o - 15o 15o
Fig. 20. The variation in the misorientation angle in dependence of changes in torsion
frequency at constans: o =6, v = 0,015 mm/s, Ah=7 mm
The Effect of Strain Path on the Microstructure and
Mechanical Properties in Cu Processed by COT Method
283
Comparing the structure of the deformed samples when the torsional frequency is lower
0,8 Hz the compression velocities are as mentioned, it is visible that the Cu samples
deformed with the higher compression velocity maintain greater fraction of the near-angle
boundaries than the samples deformed with the lower compression velocity. The
deformation realized when the compression velocity is v= 0,015 mm/s and v= 0,04 mm/s
does not cause any significant changes when it comes to the size of the grains and subgrains
(Fig.7 and Fig.18). The differences in the area fraction of the grains up to 1 m were not
observed (Fig.8 and Fig.19).
In the case of the deformation when the compression velocity was 0,04 mm/s and the
torsional frequency was 1,8 Hz, it was observed that Cu structure was significantly smaller
than after the deformation with the compression velocity equaling v= 0,015 mm/s (Fig.6d
and Fig.17b). The area fraction of the grains up to 1 m are also greater (Fig. 19) as well as
the high-angle boundaries fractions (Fig.20).
STEM micrographs (Fig.21) evidently demonstrate that deformation at higher value of v
parameter leads to generating banded structure with low angle grain boundaries and high
value of dislocation denity (Fig.22).
a) b)
c)
Fig. 21. Microstructure of Cu after COT deformation at: a) v=0,04 mm/s, b) 0,1mm/s, c) 0,6
mm/s and constans: f=1,6 Hz, o =6 and Ah=7 mm
Recrystallization
284
Fig. 22. The variation of the dislocation density In COTes samples with deferent value of
torsion frequency and constant o =6, v=0,04 mm/s and Ah=7 mm
4.3 The absolute deformation
The absolute deformation is the parameter, the increase of which has an impact on the
increase of the value of c
f
parameter. Acording with literature, the heterogeneity is
connected with too small effective deformation [Kuziak, 2005]. Using larger effective
deformation leads to the homogenization of the microstructure and larger grain refining.
Analyzing the selected samples of the size distribution of the subgrains (Fig.23), grains
(Fig.24), and the shape indicator (Fig.25) using the EBSD technique, it is seen that the
microstructure after applying a larger absolute deformation, is characterized by greater
homogeneity. The homogenous character of the microstructure is indicated by the narrow
spectrum of the size distribution in subgrains/grains and shape indicator. In the samples
which are deformed to lower values of the absolute deformation, the great diversity of grain
and subgrain size is observed.
The effect connected with the accumulation of the deformation through the increase of Ah is,
above all, the grains fragmentation by the generation of the dislocation boundaries. The
structures obtained after the absolute strain Ah =3mm, are characterized by the initial stage
of creating the DDWs boundaries (Fig.26). In the inside of the areas that are divided by the
dislocation boundaries, the cellular dislocation structure can be seen (Fig.26). After applying
the deformation Ah =7 mm, the dislocation boundaries indicate a clear contrast proving the
accumulating within the dislocation boundaries. A frequent intersection of the dislocation
boundaries is observed (Fig.27).
The Effect of Strain Path on the Microstructure and
Mechanical Properties in Cu Processed by COT Method
285
Fig. 23. Area fraction subgrains distributions of COTed Cu samples at: Ah= 3mm, Ah= 7mm;
and at constant parameters: a) f=0,8 Hz, o =6, v=0,04 mm/s , b) f=1,6 Hz, o =6, v=0,04
mm/s
Recrystallization
286
Fig. 24. Area fraction subgrains distributions of COTed Cu samples at: Ah= 3mm, Ah= 7mm;
and at constant parameters: a) f=0,8 Hz, o =6, v=0,04 mm/s , b) f=1,6 Hz, o =6, v=0,04 mm/s
Fig. 25. Shape indicator distributions of COTed Cu samples at: Ah= 3mm, Ah= 7mm; and
at constant parameters: a) f=0,8 Hz, o =6, v=0,04 mm/s , b) f=1,6 Hz, o =6, v=0,04 mm/s
The Effect of Strain Path on the Microstructure and
Mechanical Properties in Cu Processed by COT Method
287
Fig. 26. Microstructure of Cu after COT deformation at: f=1,6 Hz, o =6, v=0,04 mm/s and
Ah=7 mm
Fig. 27. Microstructure of Cu after COT deformation at: f=1,6 Hz, o =6, v=0,04 mm/s and
Ah=7 mm
The above data shows that for the effective structure refining, it would be beneficial to lead
the deformations with:
- the torsional frequency 0,8 Hz and 1,6 Hz. The large equivalent deformations realized by
the increase of the torsional frequency ranging from 0,8 Hz to 1,6 Hz, have an impact on
the increase of the misorientation value between the created grains (Fig.20), on the increase
of the area fraction of the grains with the average diameter up to 1m (Fig.19), on the
creation of the equiaxed grains and subgrains in a range of the ultrametric sizes (Fig.18).
- the high value of the absolute deformation (Ah=7 mm). The increase of the effective
deformation by means of the increase of the absolute strain has an influence on the
raise of the high-angle boundaries fraction, the increase of the defects, and the
generation of the dislocation boundaries which are subject to the mutual intersection
(Fig.26,27). The microstructures are characterized by great homogeneity (Figs.23-25).
Recrystallization
288
- the low compression velocity up to approximately 0,04 mm/s. The deformation when the
compression velocities are higher than 0,04 mm/s has no positive influence on refining a
grain to the ultrametric level (Fig.16b.c). However, the deformation proceeding when the
compression velocities are very slow, fosters the structure recovery processes (Fig.6d).
Taking into account the most convenient conditions of the Cu grain refining process it
should be stated that:
- the average diameter of the Cu grains and subgrains equals correspondingly about 300
nm and 200 nm.
- the area fraction of the grain which has the ultrafine-grained size, is 60%
- the fraction of high -angle boundaries (HAB) reaches up to about 50%
Despite the majority of the literature data does not discuss the created grain/subgrain
separately but rather give the overall values, it should be stated that the obtained are
comparable with the literature data [Dobatkin, 2007; Dalla Torre, 2004].
Less attention is paid to determine the fraction of high angle boundaries (HAB) while
describing the structural effects. Only the work of Richert [Richert, 2006] and Dobatkin
[Dobatkin, 2007] proved that the fraction of high angle boundaries (HABs) in Cu is above
50%. It is difficult to relate the obtained data concerning the area fraction of the
ultrametric grains to the literature data because the available literature does not give such
results.
The demonstrated decrease of a dislocation density with the increasing deformation,
argues for the recovery processes accompanying deformation. The literature also tells
about the decrease of the dislocation density after obtaining large deformations [Dalla
Torre, 2004].
5. Copper mechanical properties after compression with oscillatory torsion
The measurements of hardness for Cu which was subjected to compression with
oscillatory torsion, shows that the changes of hardness depends both on the compression
velocity and torsional frequency (Fig.28). The growth of the torsional frequency causes
the gradual decrease of hardness. However, the increase of the compression velocity
causes the increase of hardness. When the torsional frequency was f= 0,2 Hz, the levels of
hardness were the highest - even 130 HV0,2. The material deformed when the values of f
parameter are low, does not create the grain/subgrain structure but is only characterized
by numerous dislocation tangles. Thus, a large dislocation density is responsible for high
hardness. The growth of the effective deformations c
f
caused by the increase of the
torsional frequency does not contribute to the increase of hardness. A slow decrease of
hardness that was observed is connected with the reduction of the dislocation density
resulting from the recovering processes that it undergo. The data shown in the Fig.29,30
suggests that the refined grain of Cu leads to 1,5 fold increase of the ultimate tensile
strength (UTS) when compared with the initial stage. It was also stated that the highest
increase of the mechanical properties accompanies the deformed material when the
compression velocities are higher. In general, the increase of deformation realized
through the increase of the torsional frequency does not increase the UTS and yeld stress
(YS) of the tested materials.
The Effect of Strain Path on the Microstructure and
Mechanical Properties in Cu Processed by COT Method
289
Fig. 28. Hardnes ploted as a function of torsion frequency and compression velocity;
constants rest parameters: o =6, Ah=7 mm
Fig. 29. YS and UTS ploted as a function of torsion frequency; constants rest
parameters: v=0,015 mm/s, o =6, Ah=7 mm
Fig. 30. YS and UTS ploted as a function of torsion frequency; constants rest
parameters: v=0,04 mm/s, o =6, Ah=7 mm
Recrystallization
290
Fig. 31. A
gt
, A
c
ploted as a function of torsion frequency; constants rest parameters: v=0,015
mm/s, o =6, Ah= 7
Fig. 32. A
gt
, A
c
ploted as a function of torsion frequency; constants rest parameters: v=0,04
mm/s, o =6, Ah= 7
The reason of low ductility of the material deformed by the SPD techniques is the
localization of deformation. In the case of ultrafine-grained materials with the average
diameter of the grain not exceeding 1000 nm, the elongations are usually bigger than in
the case of nanograined materials. It is caused by the greater abilities of the material to
cumulate the dislocations in grains what leads to higher velocity of the work hardening.
The Cu plasticity expressed by the elongation to the rupture A
c
is lower in comparison to
the initial state and does not change with the variable of deformation parameters (Fig. 31
and Fig. 32). However, the total uniform elongation A
gt
reaches value of about 2% (Fig. 31
and Fig. 32).
The Effect of Strain Path on the Microstructure and
Mechanical Properties in Cu Processed by COT Method
291
(a) (b)
(c)
(a) (b)
(c)
Fig. 33. Example fracture of the specimens 4 subjected to tension, b) necking near fracture
section, c) fracture surface
Cracking of the materials after the tensile test happens in the areas of the measured parts
(Fig. 33a). The material loses its stability as a result of the deformation localization in form of
a neck which concentrates almost whole plastic deformation. In connection with that, the
uniform elongation is negligible, however, the (YS) is close to the (UTS). The fracture surface
prove the ductile character (Fig. 33b,c).
The mechanical properties of Cu after the deformation ensures the best grain size reduction
is when f=1,6 Hz, Ah=7, o=6, v=0,04 reaches values of about: YS =308 MPa and UTS = 324
MPa. The average UTS of Cu after the ECAP deformation is about 400 MPa [Kurzydowski,
2004], and sometimes even higher values were noted [Beresterci, 2008]. The plastic
properties, especially the uniform elongation for Cu, reaches the values of about 2%. This is
the value comparable with the literature data [Kurzydowski, 2004].
6. The mechanism of producing ultrafine-grained structures after the COT
deformation
A lot of place in the structural tests is devoted to the mechanisms of grain-refining with
the help of large SPD deformations. This matter is interesting particularly because of the
application of different materials and different SPD techniques. The most important
conclusions taken from the researches on the mechanisms of grain-refining are as
follows:
- The fragmentation of grain takes place when the dislocation boundaries taking
different forms are generated.
- The deformation is accompanied by the processes of dynamic recovery or even
recrystallization [Kaibyshev, 2005, Wang, 2003]. An example of a material in which the
grain-refining is the effect of deformation and of the extended recovery is aluminum.
Whereas, in copper recrystalized grains are visible in the background of the deformed
structure.
- The increase of misorientation happening thanks to the rotation of the grains
boundaries and is a diffusion- controlled process. It is based on annihilation and
absorption of the dislocation through the grains boundaries.
Recrystallization
292
Hughnes and Hansen [Hughnes, 2000] presented the concept of microstructure evolution
for the classic techniques of deformation which is based on the generation of dislocation
boundaries which lead to the division of the initial grains into smaller volumes.
The proposed conception of the structure evolution is also characteristic for SPD techniques
because many researchers dealing with refining the grain using large plastic deformations,
introduce to the description of the structure the terminology in a form of the shear bands or
dislocation layers. In general, there are three main mechanisms of the material structure-
refining that are known:
- the production of new grains takes place thanks to a gradual increase in misorientation
of dislocation boundaries as a result of absorption of new dislocations created during
the deformation [Xu, 2005],
- the fragmentation of grains takes place thanks to the generation of the shear bands
[Richert, 2006],
- the fragmentation of grains takes place thanks to the production of new grains as a
result of the continuous recovery or continuous recrystallization [Kaibyshev, 2005,
Wang, 2003] .
The refining of the copper grain after the COT deformation, happens as a result of the
generation of dislocation boundaries, which together with the growth of deformation transform
themselves into ultrafine-grained structure. The introductory stage of the grain-refining is the
creation of DDWs dislocation walls the misorientation of which reaches even a few degrees and
they stretch along the considerable fraction of a grain, separating blocks of dislocation cells
(CBs) (Fig.34). Within the boundaries a high density of dislocation is accumulated (Fig.34).
Fig. 34. Microstructure of Cu after COT deformation at: f=0,2 Hz, v=0,015 mm/s, o=6,
Ah=7 mm
The Effect of Strain Path on the Microstructure and
Mechanical Properties in Cu Processed by COT Method
293
Fig. 35. Microstructure of Cu after COT deformation at: f=0,8 Hz f=0,2 Hz, v=0,015 mm/s,
o=6, Ah=7 mm
The growth of deformation causes the transformation of the DDWs dislocation walls into
lamellar boundaries (LBs) which has larger misorientation (sometimes above 15 - HABs)
which resembles long subgrains (Fig.35) Inside of the lamellar boundaries, the dislocation
structure is regular and singular dislocation cells are usually noticeable. Moreover, the
distance between the lamellar areas decreases (Fig.35). It can be assumed that the
accumulation of deformation induces not only the creation of new dislocation boundaries
and high angle boundaries but above all, it induces the crossing of dislocation boundaries.
This phenomenon of intensive boundaries crossing (Fig.36) is a result of activating the
subsequent slip systems while the process of deformation. The places where the dislocation
boundaries are crossing induce the generation of almost equiaxed subgrains/grains (Fig.37).
As a result, the large deformation increases the share of grains/subgrains boundaries and
misorientation scattering (Fig.6) This, in turn, leads to the formation of the grains with
HABs boundaries (Fig.37). The result of EBSD test shown in the Fig.6 and Fig.17 also
constitute the confirmation of the tests. On the basis of EBSD tests it was proved that in a
great deal of cases elongated neighboring grains remain in crystallographic compatibility. In
the case of fine, equiaxed grains, the orientation was accidental (Fig.6).
The high-angle boundaries (HABs) marked in the Fig.36, 37 have bulges characteristic for
the dynamic recrystallization and they can indicate the migration of high-angle boundaries.
In many works it was proven that the migration of the grain boundaries created as a result
of SPD process as an effect of dynamic recrystallization [Kaibyshev, 2005; Wang, 2003]. The
sequence of figures registered during the rotation of the sample at a given angle indicates
that the bulges of HAB boundaries are not the effects of the boundary migration but of the
mutual superimposing of the boundaries which are in one crystallographic orientation in a
given microarea (Fig.38).
Recrystallization
294
This means that the creation of ultrafine-grained structures using the COT method is not
determined by the process of recrystallization.
Fig. 36. Microstructure of Cu after COT deformation at: f=0,8Hz, v=0,015 mm/s, o=6,
Ah=7 mm
Fig. 37. Microstructure of Cu after COT deformation at: 1,6 Hz, v=0,015 mm/s, o=6, Ah=7 mm
The Effect of Strain Path on the Microstructure and
Mechanical Properties in Cu Processed by COT Method
295
In order to trace the way in which the low-angle boundaries (LABs) change into HABs
boundaries, a series of structural tests was carried out in which the Kikuch diffraction was
used. Some examples are presented in the Fig. 39. Defining of the local orientations and
crystallographic misorientations of particular areas, allowed to formulate the mechanism in
which the high-angle boundaries are formed. The examples presented in Fig.39 suggest that
the recovery of Cu caused by EBU happens relatively slow. That is why the dislocations
generated during the deformation do not easily annihilation. The cellular type of the
dislocation structure remains even if a large deformation is used. The EBSD tests show that
the fraction of LABs boundaries is still present even after a large deformation (Fig.9,20).
Fig. 38. Microstructure of Cu after COT deformation at: 0,8 Hz, v=0,15, o=6, Ah=7 mm.
ABC marked grains are wisible after changes in sample rotation : a) o= -3,9, b) o= -6,2 , c)
o= -9,6 , d) o= -10,2 . Diffraction patterns taken from X
Recrystallization
296
Fig. 39. a) Microstructure of Cu after COT deformation at: f= 1,6 Hz, o=6, v = 0,04 mm/s.
High misorientation of nonequilibrium grain boundary in grain marked 2.; b) Kikuchy
difractions with solutions. Numbers 1-7 in Fig. 38a coresponds Kikuchy patterns 1-7 ; c)
orientation of analized area
The Effect of Strain Path on the Microstructure and
Mechanical Properties in Cu Processed by COT Method
297
The clusters of dislocation are usually created near the boundaries what causes
nonequilibrium state of the grains boundaries. A weak and heterogeneous diffractional
contrast inside of the grains indicates a high level of the interior stress in the grain
boundaries (Fig.39). Such boundaries are high-angular but almost invisible.
The structural analysis presented here show that the dominating mechanism of forming
the high-angle grains in Cu is the growth of misorientation in dislocation boundaries.
This, in turn, happens as an effect of absorbing dislocations to the grains boundaries
during the deformation process. The nonequilibrium boundaries presented in the Fig.39
are formed from the absorption of numerous dislocations as a result of glide and climb
[32].
It was proven that together with the growth of deformation, the fraction of the
boundaries increases gradually from 5-15 (up to approx. 30%) and the fraction of HABs
boundaries raises above15 (to about 40%) (Fig.6). Approximately, on a constant level the
fraction of narrow-angle boundaries LABs is formed and it does not exceed 30%. When
the process parameters are changed (Fig.20) the increase of the HABs fraction to 50% is
visible, but also the fraction of boundaries form 5-15 was reduced to approximately
20%.
The LABs boundaries still constituted 30%. It can mean that the boundaries with
misorientation from 5-15 transform themselves into the grain boundaries. Because of the
fact that the boundaries with an average misorientation (5-15) are the nonstable
boundaries [Cao, 2008], as a result of the deformation they are transformed into high-angle
boundaries.
7. Summary
The chapter concentrates on the possibility of a structure forming and on the properties of
the metallic materials by means of using the method of compression with oscillatory torsion.
The issues that were introduced have a great technical importance because it refers to the
formation of ultrafine-grained structure by application of the unfamiliar SPD technique.
Because the intensive research on generating new, more economical methods of producing
ultrafine-grained materials, the presented matter coincides with the current state of research
conducted in many scientific centers.
The aim of the tests in which the COT deformation method was used, was to obtain
maximal refining of the grain below 1m. The structural investigations showed that
using the COT method, the grains of the Cu can be refined to ultrafine-grained.
The intensity of grain-refining depends on the value of the effective deformation,
however, the deformation path (the selection of the deformation parameters) is a decisive
factor.
Using the different deformation parameters in process the presence of different phenomena
that were controlling the growth of the microstructure. The increase of deformation
(realized through the Hz growth) causes a progress in the grain refinement. The
deformation conducted when the torsional frequencies were 0,8 Hz and 1,6 Hz is the most
beneficial for obtaining the most refined grain. Using a significantly higher torsional
Recrystallization
298
frequency -18 Hz during the deformation caused considerable restrictions in the grain-
refining because of the intensive recovery which begins to dominate over the deformation
process.
It can be noticed that a relatively small change of the compression velocity, had an
impact on considerably greater refining of the structure. The acceleration of the
deformation process by increasing the compression velocity, causes the delay of the
structure recovery process. What is seen through the changes of dislocation density for
example. Despite reducing the effective deformation, the progress in the structure
refining is observed what denies the results described in the literature [Dalla Tore, 2004;
Wang, 2003]. However, the next increase of the compression velocity from 0,1 mm/s and
0,6 mm/s does not foster grain-refining. It should be explained by the fact that the
effective deformation is too small.
The main effect of deformation is the increase of the structures homogeneity. The
homogeneity is obtained mainly by the increase of the height parameter Ah.
When the process parameters are as follows: the torsional frequency ranges from 0,8 Hz to
1,6 Hz; Ah=7mm, and the compression velocity ranges from 0,015 to 0,04 mm/s; the
maximal refining of the copper grain is obtained but also:
- the average diameter of the grain/subgrain correspondingly 200 nm and 300 nm.
- the share of high-angle boundaries is about 50% and
- the fraction of the ultrafine grains is about 60%. In spite of using large equivalent
deformations, narrow-angle boundaries up to 15 constitute a significant fraction in
both materials. The fraction of high-angle boundaries does not exceed 70%. This means
that the thermal stability of the structures formed in such a way is sufficient [Raab,
2004; Lugo, 2008].
The result of mechanical tests were discussed in details for selected schemes of deformation
in which the effects of deformation were visible the most. The influence of grain refining on
the mechanical properties of the grain, was determined on the basis of the hardness
measurements and the tensile test of the micro samples. The results of the mechanical tests
that were carried out, indicate clearly that the changes taking place in the structure and the
grain refining had a positive influence on the mechanical properties of the tested materials.
However, the uniform elongation is not strongly dependent on the grain refining.
The difference in the properties that were observed, mainly result from using different
schemes of deformation.
The results of structural investigations were the basis for determining the mechanisms of
grain refining and forming high-angle boundaries. It was proven that the formation of
ultrafine grains during the COT deformation is based on the general mechanisms of
forming the dislocation boundaries. The process of grain refining proceeds by the
generation of the LABs and HABs boundaries. In the introductory stage of deformation
the dislocation boundaries are formed which are perpendicular to the direction of the
compression force. The formation of the dislocation boundaries which proceed in such a
way, suggests that in the initial stage of deformation it is mainly the compression that
initiates the process of the grain refining. The non-directional process of deformation
The Effect of Strain Path on the Microstructure and
Mechanical Properties in Cu Processed by COT Method
299
(the introduction of an additional torsion causing the change in the direction of loading)
leads to the deformation of the material in more and more numerous systems of glides
and to the new dislocations having influence on the active system of glides on the
previously-generated dislocations. The effect of the introduced, additional loading is the
increase in the number of the dislocation boundaries that cross mutually. When the
effective deformation in the microstructure increases, the distances between the
dislocation boundaries decrease a new order of LBs dislocation boundaries is created
which as a result of the thermal process activation (recovery), transform themselves into
subgrains and grains. The process leads to the increase of misorientation of the grains
boundaries.
A significant role in forming the ultrafine-grained structure has the recovery process.
Dislocations are rearranged, undergo annihilation and are also absorbed to the grain
boundaries. Such a rebuilding of a dislocation structure causes the increase of the
misorientation within the grain boundaries.
8. References
Pakiea, Z. (2009). Mikrostrukturalne uwarunkowania waciwoci mechanicznych
nanokrystalicznych metali, Oficyna Wydawnicza Politechniki Warszawskiej
Olejnik, L. et al. (2005). Bulletin the polish academy of science, Technical sciences, 53, pp. 413-423
Bochniak, W. et al. (2005). Journal of Materials Processing Technology, 169, pp. 44-53
Shaarbaf, M. et al. (2008). Materials Science Engineering, A473, pp. 28-33
Raab, G.J. et al. (2004). Mater. Sci. Eng., A 382, pp. 30
Lugo, A. N. et al. (2008). Materials Science Engineering, A477, pp. 366-371
Kulczyk, M. et al. (2007). Materials Science Poland, 25, pp. 991-999
Pawlicki J.& F. Grosman. (2005). Naprenia uplastyczniajce metali w warunkach
zoonych obcie, Materiay XII Konferencji Informatyka w Technologii Metali,
Ustro 16-19.01
Grosman F. & Pawlicki J. (2004). Processes with forced deformation path. New Forming
Technology 2004. Proceedings of the 1
st
ICNFT, Harbin Institute of Technology Press,
Harbin, China, Sep. 69
Rodak, K. & Goryczka T. (2007). Solid State Phenom, 130, pp. 111-113
Ungar, T. et al. (2001). Materials Science Engineering, A319-321, pp. 274-278
Kuziak, R.(2005). Modelowanie zmian struktury i przemian fazowych zachodzcych w procesach
obrbki cieplno-plastycznej stali, Instytut Metalurgii elaza, Gliwice
Dobatkin, S. V. et al. (2007). Materials Science Engineering, A462, pp. 132-138
Dalla Torre, F. (2004), Acta Materialia, 52, pp. 4819- 4832
Richert, M. (2006). Inynieria nanomateriaw i struktur ultra drobnoziarnistych, Uczelniane
Wydawnictwo Naukowo Techniczne, Krakw
Kurzydowski, K. J. (2004). Biuletin of the polish academy of sciences technical sciences, 52, pp.
301-311
Beresterci, M. et al. (2008). Metalurgija, 47, pp. 295-299
Kaibyshev, R. et al. (2005). Materials Science Engineering, A398, pp. 341-351
Wang, G. et al. (2003). Materials Science Engineering, A346, pp. 83-90
Hughnes, D.A. et al. (2000). Acta Materialia, 48, pp. 2985-3004
Recrystallization
300
Xu, Ch. et al. (2005). Materials Science Engineering, A398, pp. 66-76
Cao, W.Q. et al. (2008). Materials Science Engineering, A492, pp. 74-79
Part 2
Recrystallization of Minerals
12
Zircon Recrystallization History as a Function of
the U-Content and Its Geochronologic
Implications: Empirical Facts on Zircons from
Romanian Carpathians and Dobrogea
Ioan Coriolan Balintoni and Constantin Balica
Babe-Bolyai University, Cluj-Napoca
Romania
1. Introduction
According to Davies et al. (2003), "the age of Earth and the time scale of pre-human events
are central to a civilization's sense of origin and purpose. Therefore, the quest for precise
and reliable geochronometers has had a scientific and cultural importance that few other
enterprises can match". In this respect, since the beginning of the last century it has been
recognized that long-lived radioactive decay systems provide the only valid means of
quantifying geologic time.
One of the most reliable Earth's timekeeper has proven to be the mineral zircon, since it
records the ages of Earths earliest evolution stages, the oldest sediments, extinction
episodes, mountain-building events and supercontinents coalescence and dispersal (e.g.
Rubatto and Hermann, 2007; Harley et al., 2007). Its widespread use in geochronology is
based on the decay of uranium (U) and thorium (Th) to lead (Pb). They provide three
distinct radioactive decay series involving the parent isotopes
238
U,
235
U and
232
Th and their
daughter isotopes
206
Pb,
207
Pb and
208
Pb, respectively. Through the incorporation of U and
Th at the time of growth, every zircon grain hosts three different clocks. In an ideal closed
system, the three estimates would agree with each other within the analytical errors of
measurements. However, in natural systems the zircon grains are not equally closed for Th
and U with respect to post-crystallization effects. The usual approach in zircon
geochronology is to consider the U-Pb system alone, as there is no natural non-nuclear ways
of fractionating
235
U from
238
U. As the modern day-ratio of
235
U/
238
U is well known
(1/137.88) the need to actually analyze very low abundances of
235
U is obviated. Besides U
and Th, zircon can incorporate some other incompatible elements such as P, Sc, Nb, Hf, Ti,
and REE in trace (up to thousands of ppm) or minor (up to 3% wt) amounts. The primary
control factor on the substitutions is the ionic radii of the substituting cations compared with
Zr
4+
and Si
4+
cations. Substitutions that minimize strain effects on either or both sites will be
favored. The crystal-chemical limitations are that Zr
4+
in 8-fold coordination has an ionic
radius of 84*10
-3
nm and Si
4+
, in tetrahedral coordination, has an ionic radius of 26*10
-3
nm.
U
4+
(ionic radius of 10*10
-2
nm in 8-fold coordination) and Th (105*10
-3
nm in 8-fold
coordination) can be accommodated in the Zr
4+
sites. Uranium concentrations are usually
Recrystallization
304
less than 5000 ppm and Th concentration less than 1000 ppm. Because of its ionic radius of
129*10
-3
nm (8-fold coordination), Pb
2+
is highly incompatible with growing zircon crystal
lattice and therefore is not incorporated more than ppb levels, which is crucial in
geochronology. Because of the same reason, Pb
2+
can easily escape from zircon lattice when
some conditions are fulfilled.
Based on idea of concordant ages between
235
U/
207
Pb and
238
U/
206
Pb, Wetherill (1956) has
developed the Concordia diagram. Quite soon though it has been shown (Tilton et al., 1957)
that the concordance situations are rather rare and usually zircon shows evidence of
discordance (disagreement between
235
U/
237
Pb and
238
U/
206
Pb ages) due to Pb loss caused
by some post-crystallization geologic events.
Since then, the attempts made for understanding the causes of discordance had become the
main preoccupation of the U-Pb geochronologists over the next quarter of century. Finally, it
has been understood that discordance can be attributed to two major causes: (1) - mixing, in
the analyzed sample volume, of discrete zones of different ages from within the same zircon
grain; (2) - partial loss of radiogenic Pb by the entire zircon grain or by fractions of it. While
the former cause can be bypassed by in situ dating, the second still hinders unequivocal
results.
Continuously growing database of published in situ age data provided further insights into
the behavior of U-Pb system in zircons: (1) - radiogenic Pb can be entirely lost and even the
concordant data do not indicate the initial crystallization age of zircon; (2) frequently, Pb
loss is accompanied by Th loss; (3) occasionally, even the U can be lost.
The disturbance of the isotope systematics in zircon is related to: (1) - amorphization; (2) -
alteration; (3) - recrystallization.
The focus of the present contribution is on the discussion of the above three processes
exemplified with their geochronological consequences on zircons from the Romanian
Carpathians and Dobrogea. All data presented here were obtained by in situ dating through
Laser Ablation-Inductively Coupled Plasma-Mass Specterometry (LA ICP-MS).
1.1 Amorphization process
This process is due to -decay events associated with U and Th radioactive disintegration.
Based on X-ray diffraction and High Resolution Transmited Electron Microscopy (HRTEM)
analysis, Murakami et al. (1991) suggested three stages of damage accumulation in Sri
Lankan zircon: in stage I (at <3 x 10
18
-decay events/g) the accumulation of isolated point
defects predominate. These defects have the potential to recover through geologic time;
stage II (at 3x10
18
to 8x10
18
-decay events/g) is evidenced by crystalline regions with point
defects and amorphous tracks caused by overlapped -recoil nuclei; during stage III (at >
8x10
18
-decay events/g) only aperiodic material can be screened by X-ray and electron
diffraction.
Salje et al. (1999) suggested a two-phase transition during increasing amorphization
process. The first phase is related to the percolation of amorphous material into the
crystalline matrix and the second one to the percolation of crystalline material into the
amorphous matrix.
Zircon Recrystallization History as a Function of the U-Content and Its Geochronologic
Implications: Empirical Facts on Zircons from Romanian Carpathians and Dobrogea
305
According to Nasdala et al. (2001) the -decay events in the decay chains of U and Th cause
the zircon amorphization. An -particle generates about 120-130 Frenkel type defect pairs
along penetration distances from 9.9 to 29.5m. Recoils of heavy daughter nuclei are only a
few hundred in length but the recoil damage clusters include 600-1200 Frenkel defect
pairs. Spontaneous fission fragments produce heavy damage locally, but their contribution
to the overall radiation damage is of minor importance because of their relative rarity. In the
absence of recovery, radiation damage is stored in zircon, causing transformation from the
crystalline to the metamict state. These authors propose the following stages of radiation
damage accumulation.
1. Scattered nano-regions with high defect concentration. The amorphous component is
still insignificant.
2. Moderately radiation damaged zircon in which amorphous nano-regions form a
domain structure. Amorphous domains in a crystalline zircon or crystalline remnants in
an amorphous matrix can be observed.
3. Entirely aperiodic zircon.
Ewing et al. (2003) speak about -particles ionization processes over a range of 16 to 22
microns that produce several hundreds isolated atomic displacements. The associated -
recoils lose their energy during elastic collisions over 30 to 40 nm, producing localized
collision cascades of 1000 to 2000 displacements. However, with increasing temperature the
amorphization dose increases. This can be due either, by a decrease in the average cascade
size caused by thermal relaxation or by a reduction in the surviving amorphous volume as a
result of thermal recovery of irradiation induced defects.
According to Geisler et al. (2007), the amorphization of a crystalline zircon is characterized
by transformation from an initial stage, where isolated amorphous domains are surrounded
by slightly disordered crystalline material, to a more advanced stage of damage where few
isolated, disordered, nano-crystalline islands occur in an amorphous matrix. These authors
refer to Salje et al. (1999) when describing the crystalline- to-amorphous transformation as a
geometrical phase transition. During the first phase transition, the amorphous domains
form clusters that percolate over macroscopic length scales, whereas during the second
transition, the crystalline domains cease to be interconnected. The first percolation point
appears at an amorphous fraction of ~30% while the second percolation point appears when
the amorphous fraction reaches ~ 70%. In other words, less than 30% amorphous fraction
and less than 30% crystalline fraction can not be interconnected.
1.2 Alteration processes
According to Geisler et al. (2002), alteration is the interaction between the metamict zircon
and fluids, including the meteoric ones (weathering processes), characterized by distinct
chemical and structural changes. Geisler et al. (2003) describe two anomalous stages in the
alteration rate with increasing degree of amorphization. The first stage takes place when the
amorphous domains form interconnected clusters within the zircon structure, namely at the
first percolation point suggested by Salje et al. (1999). At this point, the percolation interfaces
that represent low-density areas between crystalline and amorphous domains open high
diffusivity pathways. A new dramatic increase in alteration rate is observed at the second
percolation point of Salje et al. (1999). Around this point, a network of nanometer-size
Recrystallization
306
regions of depleted matter interconnects discrete amorphous domains without crystalline
dams.
According to Geisler et al. (2007), the structural changes that take place in a zircon grain are
defined by the fact that above the 200
o
C limit the thermal recrystallization front generated
by epitaxial reordering moves inward the zircon crystal. Bellow the 200
o
C threshold, the
thermal recrystallization involves enhanced defect diffusion only along the reordering front.
The above model hypothesizes an increase in the effective diffusion of any species within
the zircon lattice with increasing -decay dose or of amorphization degree.
1.3 Recrystallization
There are several types of recrystallization processes that take place in zircons and several
different opinions in terms of its significance exist on this matter.
According to Nasdala et al. (2001), recrystallization is a "re-growth" process in the
crystallographic sense. During recrystallization a new zircon lattice forms along a
crystallization front, typically replacing a more disordered and polluted zircon.
Recrystallization leads to healing of the radiation damage and partial or complete resetting
of the U-Pb isotopic system. According to the same authors, annealing requires only re-
formation of disrupted bonds by re-ordering of nearest neighboring atoms, thus annealing
of zircon structure is not necessarily associated with any Pb loss. As defined in the literature,
recrystallization presumes an epitaxial migration of an interface between an ordered region
in zircon and a metamict vicinity, therefore it could be described as a defect annihilation or
point defect diffusion process. In a general view, Nasdala et al. (2001) enumerate the
following thermal recovery mechanisms: (1) - point defect diffusion in the crystalline and
amorphous phase; (2) - epitaxial growth of crystalline residuals; (3) - random nucleation in
the amorphous phase.
Ewing et al. (2003) put forward a more comprehensive description of the concept of
recrystallization, while offering the foundation for a clear distinction between Type I and
TypeII recrystallization. Type I recrystallization is purely thermal and occurs on time scales
longer than cascade quench time. This behavior is due to point defect diffusion and epitaxial
migration of the crystalline residuals toward the amorphous domains. It can be seen that
Type I recrystallization, as described by these authors, covers both the recrystallization and
annealing processes of Nasdala et al. (2001). Type I recrystallization prevails over longer
periods of time or at higher temperatures and becomes particularly important in natural
specimens stored at ambient conditions for geological periods. In moderately damaged
zircons, two stages of recovery process have been described. The first stage is defined by the
recovery of short-range order and point defect recombination. It occurs bellow ~ 600
o
C
(Farges, 1994) or ~727
o
C (Geisler et al., 2001). The second stage occurs at higher
temperatures and is caused by epitaxial recrystallization along the internal crystalline-
amorphous boundaries. It is worth noting that an initially moderately damaged zircon grain
consists of distorted crystalline phases embedded in an aperiodic matrix at the completion
of the first phase. In contrast to the moderately damaged specimens, heavily damaged or
amorphous zircon segregates into its constituent oxides at higher temperatures and
recrystallization takes place in three stages: (1) - decomposition of amorphous zircon into
tetragonal ZrO
2
crystallites and amorphous SiO
2
; (2) - tetragonal to monoclinic phase
Zircon Recrystallization History as a Function of the U-Content and Its Geochronologic
Implications: Empirical Facts on Zircons from Romanian Carpathians and Dobrogea
307
transformation in the ZrO
2
crystallites; (3) - formation of coarse-grained (several hundred
m) randomly oriented, polycrystalline zircon. This process is highly improbable to take
place in natural zircon and it excludes the idea of new zircon nucleation within amorphous
phase as postulated by Nasdala et al. (2001).
Type II recrystallization occurs as a nearly instantaneous process during irradiation and can
be divided into two distinct phases. The first, Type IIa recrystallization, is due to increased
mobility of interstitials and other point defects during irradiation. The irradiation enhanced
diffusion leads to a greater degree of point defect recombination and annihilation. Point
defect annihilation is most effective at structural boundaries between amorphous and
damaged, but still crystalline regions. Type IIb recrystallization occurs within individual
displacement cascades. Disordered and highly energetic material can epitaxially
recrystallize at the cascade peripheries along with the cooling of the displacement cascade to
the ambient temperature. During irradiation, both Type I and Type II recrystallization
processes contribute to the dynamic recovery of zircon, but some mechanisms can prevail
over others in certain temperature regimes. Obviously, at the Earth's surface conditions,
both Type I and Type II recrystallization processes are less effective than the irradiation
damage over an -dose range, allowing amorphization accumulation in time.
Geisler et al. (2007) proposed two more kinds of zircon recrystallization, fluid or melt
assisted. The Hf, U and Th content of zircon can be also distributed in solid solutions
between zircon and hafnon (HfSiO
4
), coffinite (USiO
4
), thorite (ThSiO
4
). Because solid-state
exsolution structures have not been yet reported in zircon, it is argued that such solid
solutions are metastable after cooling and characterized by structural strain. Such structural
strain enhances surface reactivity and thus the dissolution rate. We note that the effects of
the structural strain are supplementary added to irradiation damage, increasing the
reactivity between the zircon and fluids. It was shown (Geisler et al., 2003) that the
treatment of radiation-damaged zircon crystals in various aqueous solutions produces
inward-penetrating, irregular, and curved reaction domains that resemble those found in
natural zircon. Recrystallization of zircon on the expense of the amorphous phase inside the
reacted domains occur at experimental temperatures above 200
o
C. Recrystallization of
amorphous zircon dramatically reduces the molar volume of the reacted areas inducing a
strain that is partially released by fracturing. Porous structure at nanometer-scale level is
also very likely to occur. The nanoporosity created between the crystallites provides
pathways for chemical exchange between reaction front and the fluids. This is the diffusion-
reaction process in which a moving recrystallization front follows at some distance behind
the percolation-controlled, inward diffusion of a hydrous species. According to Geisler et al.
(2007), a prerequisite for the diffusion-reaction process is the presence of more than 30 %
amorphous fraction, which is of an interconnected amorphous network.
The coupled dissolution-reprecipitation is a process by which the breaking of the bonds
and dissolution is accompanied by contemporaneous nucleation and precipitation of new
zircon. The coupled dissolution-reprecipitation process is independent of the absolute
solubility of zircon in natural aqueous fluids, which is very low (Tromans, 2006). This can
result in a complete replacement of one zircon crystal by a new one within the same
space, without losing the external shape or crystal morphology (Putnis, 2002; Putnis et al.,
2005). The chemical exchange between the dissolution-reprecipitation front and external
fluid is maintained by the formation of porosity. This porosity results from the higher
Recrystallization
308
solubility and the higher molar volume of the dissolved parent zircon, as compared with
the more pure zircon, chemically reprecipitated . Since the dissolution of a metastable
zircon solid solution is kinetically favored by structural strain, the radiation damage may
also enhance kinetically a coupled dissolution-reprecipitation process. Considering the
two proposed mechanisms, the reaction of zircon with fluids and melts provides an
effective way of its re-equilibration.
1.4 Temperature conditions of zircon recrystallization
According to Mezger and Krogstad (1997), the 600-650
G T J
P
T
(1)
From the thermodynamic equation of reaction (1), it can be seen that decomposition
temperature of Fe
2
O
3
increases with increasing oxygen partial pressure. Therefore,
Recrystallization
330
excessively high firing temperature and low oxygen partial pressure should be avoided to
restrain the decomposition of Fe
2
O
3
. Thus, it is necessary to maintain at higher roasting
temperature for hematite pellet as well as narrower firing temperature range, which makes
the operation of firing equipments difficult.
To enhance the induration of hematite pellets, both magnetite-addition and carbon-
burdened methods are found to be the favourable techniques in practice. In this chapter, the
induration mechanisms of hematite pellet with addition of magnetite concentrate and
anthracite powder are revealed by characterization of recrystallization rules of Fe
2
O
3
during
the oxidization roasting.
2. Fe
2
O
3
recrystallization during the firing of mixed hematite/magnetite
concentrates pellet
2.1 Materials and methods
2.1.1 Materials
The chemical compositions of iron ore materials and bentonite are shown in Table 1. The
size distribution of iron concentrates is shown in Table 2.
Materials Fe
total
FeO SiO
2
CaO MgO Al
2
O
3
Hematite 67.60 0.72 1.55 0.15 0.17 1.16
Magnetite 69.31 27.88 1.32 0.23 0.55 1.02
Bentonite 4.38 / 59.05 0.68 1.73 18.72
Table 1. Chemical compositions of materials /%
Materials +0.075mm 0.0375-0.075mm -0.0375mm
Hematite 13.50 16.35 70.15
Magnetite 5.93 15.35 78.72
Table 2. Size distribution of iron ore materials /%
2.1.2 Methods
The experimental procedure includes ball preparation, preheating and roasting tests,
strength measurement and mineralogical analysis.
For each trial, 5 kg mixed concentrates at the given hematite/magnetite (H/M) ratio was
blended with 8% moisture and 0.5% bentonite was used as binder. The green balls were
prepared in a disc pelletizer with a diameter of 1000 mm. The green balls of 10-12 mm in
diameter were statically dried at 105C in an electrical furnace for 4 hrs.
Preheating and roasting tests were carried out in an electrically heated horizontal tube
furnace with an internal and external diameter of 50 mm and 70 mm respectively. Firstly,
Recrystallization of Fe
2
O
3
During the Induration of Iron Ore Oxidation Pellets
331
the dry balls were put into a corundum crucible and pushed into the preheating zone of the
furnace step by step, preheated at the given temperature for a given period. Then the
preheated pellets were taken out of the furnace and cooled in the air, or directly pushed
forwards into a higher temperature zone for roasting. Finally, the roasted pellets were taken
out and naturally cooled in the air.
The compression strength of cooled pellets was measured with an LJ-1000 material
experimental machine. An average value of 20 pellets is expressed as the compression
strength for each test.
2.2 Effects of H/M ratio on the compression strength of pellet
According to the orthogonal experimental results, the relationship between H/M ratio in
pellet and the compression strength of preheated and roasted pellet was investigated and
shown in Fig. 1. Experimental conditions are preheating temperature of 900C and
preheating time of 10 min for Fig. 1a, and 900C preheating temperature, 10 min preheating
time, 1275C roasting temperature and 15 min roasting time for Fig. 1b.
It can be seen from Fig. 1a, that the preheated pellet strength is continuously improved from
190 N/P to 1132 N/P when magnetite ratio is increased from 0 to 100%. The main reason is
that magnetite has fine particle size and great specific surface area, and magnetite particles
are rapidly oxidized into Fe
2
O
3
grains during the preheated stage; moreover, the atoms on
the newborn Fe
2
O
3
grain surface have greater migrating capability than those on the original
hematite grains, and consequently the Fe
2
O
3
crystallites are easily formed between particles.
It also can be found that the compression strength of roasted pellet increases with magnetite
ratio below 70%; above 70%, the strength will be decreased with the addition of magnetite.
This may be due to the fact that some of magnetite particles are residual in the core of pellet
during the preheating stage, which cannot be completely oxidized in the high temperature
roasting stage; the magnetite particles are recrystallized and bonded with the formation of
slag phases.
2.3 Firing properties of mixed H/M pellet
2.3.1 Preheating characteristics
2.3.1.1 Compression strength of preheated pellet
The effects of preheating parameters on the compression strength of preheated pellets
with different H/M ratios are shown in Fig. 2. It can be seen from Fig. 2a that the
strength of pellet with H/M=70:30 reaches 559 N/P when preheated at 900C for 10 min,
while the strength of pellet with H/M=50:50 exceeds 560 N/P when the time is more than
5 min. Fig. 2b shows that the preheated pellet strength keeps increasing with the
preheating temperature varying from 800 to 1000C. The strength of the pellets with
H/M=50:50 reaches 470 N/P when the temperature is 800C. However, the compression
strength of the preheated pellets with H/M=70:30 is less than 400 N/P until the
temperature is over 850C. Thus, the results show that the preheating time can be
shortened and preheating temperature can be decreased with the increase of magnetite
ratio in the pellet.
Recrystallization
332
0 20 40 60 80 100
200
400
600
800
1000
1200
C
o
m
p
r
e
s
s
i
o
n
s
t
r
e
n
g
t
h
o
f
p
e
l
l
e
t
(
N
/
P
)
Magnetite ratio in pellet (%)
3000
3500
4000
4500
5000
a-Roasted pellet
b-Preheated pellet
Fig. 1. Effects of magnetite ratio on the compression strength of pellet
2.3.1.2 FeO content of preheated pellet
Whether the magnetite in pellet is oxidized completely or not has an important effect on the
roasted pellet performance (compression strength, microstructure, mineral composition,
etc.). Generally speaking, residual FeO content in the preheated pellet is required to be less
than 3% in operation. The FeO content in pellets with H/M=70:30 under different
preheating conditions were analyzed and are shown in Fig. 3. FeO content falls gradually
with the increase in preheating temperature and time. FeO content is less than 3% only if the
preheating time is more than 11 min at 850C; however, at 900C, FeO content goes down to
2.97% when the preheating time is 8 min. In consideration of the compression strength and
FeO content together, the suitable preheating parameters for the pellet with H/M=70:30
should be 8~11 min at 850C~900C.
Recrystallization of Fe
2
O
3
During the Induration of Iron Ore Oxidation Pellets
333
4 6 8 10 12 14 16
400
600
800
1000
1200
1400
a-preheating temp.: 900
o
C
H/M=70:30
H/M=50:50
C
o
m
p
r
e
s
s
i
o
n
s
t
r
e
n
g
t
h
o
f
p
r
e
h
e
a
t
e
d
p
e
l
l
e
t
(
N
/
P
)
Preheating time (min)
800 850 900 950 1000
200
400
600
800
1000
1200
b-preheating time: 10min
Preheating temperature()
Fig. 2. Effects of preheating on the compression strength of the preheated pellets with
different H/M ratios
4 6 8 10 12 14 16
1
2
3
4
5
6
7
Preheating temperature ( )
800
850
900
950
1000
F
e
O
c
o
n
t
e
n
t
o
f
p
r
e
h
e
a
t
e
d
p
e
l
l
e
t
(
%
)
Preheating time ( min)
Fig. 3. Effects of preheating on FeO content of the preheated pellets with H/M=70:30
Recrystallization
334
2.3.2 Roasting characteristics
The effects of preheating and roasting parameters on the compression strength of the
finished pellets with different H/M ratios are presented in Fig. 4.
800 850 900 950 1000
3000
3500
4000
4500
5000
H/M=70:30
H/M=50:50
d
Preheating temperature()
4 6 8 10 12 14 16
4000
4500
5000
c
C
o
m
p
r
e
s
s
i
o
n
s
t
r
e
n
g
t
h
o
f
r
o
a
s
t
e
d
p
e
l
l
e
t
(
N
/
P
)
Preheating time(min)
1200 1250 1300 1350
2000
3000
4000
5000
b
Roasting temperature ()
10 12 14 16 18 20
4000
4500
5000
a
Roasting time(min)
a-preheating time 10 min, roasting temp. 1275C, roasting time 15 min;
b-preheating temp. 900C, roasting temp. 1275C, roasting time 15 min;
c-preheating temp. 900C, preheating time 10 min, roasting time 15 min;
d-preheating temp. 900C, preheating time 10 min, roasting temp. 1275C
Fig. 4. Effects of preheating and roasting parameters on the compression strength of the
finished pellet with different H/M ratios
Recrystallization of Fe
2
O
3
During the Induration of Iron Ore Oxidation Pellets
335
As shown in Fig. 4a, the compression strength of roasted pellets increases quickly with the
preheating temperatures varying from 800 to 900C and remains almost unchanged when
temperature is between 900 and 950C. While the temperature is further increased, the pellet
strength decreases slightly. The main reason is that a compact layer is quickly formed and
the residual magnetite in the core of pellet is difficult to be oxidized completely. If pellet is
roasted in poor oxidative atmosphere at relatively high temperature, low melting point slag
phases (such as 2FeOSiO
2
) are formed, and result in deteriorating the compression
strength of roasted pellet. From Fig. 4b, the strength of the roasted pellets increased
gradually with increasing preheating time, the reason for which is that magnetite particles
are gradually oxidized into Fe
2
O
3
crystallites and they are recrystallized more completely
with the prolonged preheating time.
As shown in Fig. 4c, the pellet strength increases steadily with the increase of roasting
temperature below 1300C. The possible reason is that the Fe
3
O
4
is gradually oxidized into
Fe
2
O
3
, which leads to Fe
3+
diffusion, rearrangement of Fe
2
O
3
crystal lattices and a compact
microstructure formed. When the roasting temperature is over 1300C, however, the pellet
strength is reduced with rising temperature, the main reason for which is that: on one hand,
it is too late for the residual magnetite in the core of preheated pellet to be oxidized
adequately at high temperature, so that low melting point slag phases are formed priorly;
and on the other hand, part of Fe
2
O
3
crystal grains decomposes at high temperature
reversely, so that the structure of the pellet is destroyed to some degree. As far as the
roasting time is concerned, as given in Fig. 4d, the pellet strength is significantly improved
with the roasting time in the range of 1015 min and remains constant after 15 min.
The results mentioned above show that adding a proper proportion of magnetite
concentrate into the hematite pellet is able to improve the compression strength of both
preheated and roasted pellets.
2.4 Induration mechanisms of mixed H/M pellet
Mineral composition, microstructure and Fe
2
O
3
crystallization of fired pellets at various
preheating and roasting temperatures were studied by using Leica DMRXE microscope with
an automatic image analyzer.
2.4.1 Crystallization behavior of Fe
2
O
3
during preheating
Oxidation of magnetite into hematite is the main reaction during preheating. Oxygen
readily diffuses into the interior of the porous pellet and reacts with magnetite particles
during preheating. Oxidation always occurs firstly on the surface of particles and cracks,
and usually a few small spotty or lamellae hematite are formed (Figs. 5, 6). The oxidation
process will advance towards the core with increasing temperature.
It is shown that the hematite in the pellets involves two types: the one is original hematite
(OH), which is the unreacted hematite from the raw hematite concentrate (particle A), and
the other, namely secondary hematite (SH), is formed from the oxidation of magnetite
concentrate (particle B).
Obvious differences in colour and shape can be observed between the OH and SH grains.
For the OH grains, their colours are bright and white, and their shapes are original and with
Recrystallization
336
discernible angularity. The strip, triangular and rectangular grains can be distinctly
observed, and the surface of crystal grain is smooth and the compact inner structure is
unchanged. Moreover, the distance between the two close OH particles is large; there is no
crystallitic bond formation between them.
1-SH (columnar); 2-OH; 3-residual magnetite (irregular, tabular)
Fig. 5. Shapes of SH and OH grains in the pellet preheated at 950C
1-SH (reticular); 2-OH; 3-residual magnetite (vein)
Fig. 6. Shapes of SH and OH grains in the pellet preheated at 1000C
Recrystallization of Fe
2
O
3
During the Induration of Iron Ore Oxidation Pellets
337
By contrast, for the SH grains, they are relatively heavy-coloured, and some residual
magnetite domains can be observed. Tabular or massive grains are the main morphology of
residual Fe
3
O
4
preheated at 950C (Fig.5), but the grains change into reticular and vein
shape at 1000C (Fig. 6).
In comparison with OH grains, the angularities of the SH particles disappear or become
unclear during oxidation, and they transform into massive, zonal or columnar-shaped
particles. The formation of Fe
2
O
3
microcrystalline junctions between the close SH particles is
significantly different from the OH particles at this stage.
The above results suggest that Fe
2
O
3
in SH particles is more active than that in the OH
particles. Fe
2
O
3
microcrystalline junctions between SH particles are formed when the pellet
is preheated, which is able to improve the strength of the preheated pellet. The OH particles
keep their original shapes, and no Fe
2
O
3
microcrystalline junction can be observed at the
preheating stage, thus OH has little contribution to the strength of preheated pellets.
Therefore, the improvement of oxidative atmosphere during preheating is able to enhance
Fe
2
O
3
microcrystalline junctions between the SH particles, the strength of preheated pellet
will be improved accordingly.
2.4.2 Crystallization behaviour of Fe
2
O
3
during roasting
The crystal morphology of Fe
2
O
3
and microstructure of pellets roasted at different
temperatures are shown in Figs. 79.
1-SH particle; 2-OH particle
Fig. 7. OH grains enclosed by or connected with SH grains in the pellet roasted at 1150C
As shown in Fig. 7, a large number of junctions between particles are formed by
recrystallization of Fe
2
O
3
at 1150C and the strength of the roasted pellet highly increases.
However, the inner structure and shape of the OH particles remain visible, which indicates
Recrystallization
338
that none of Fe
2
O
3
recrystallization takes place in the OH particles at 1150C. The junctions
between particles come from Fe
2
O
3
crystallization of the SH particles.
1-SH particle; 2-OH particle
Fig. 8. Recrystallization of Fe
2
O
3
in the pellet roasted at 1230C
Fig. 9. Developed Fe
2
O
3
recrystallization in the pellet roasted at 1280C
Recrystallization of Fe
2
O
3
During the Induration of Iron Ore Oxidation Pellets
339
At 1230C (Fig. 8), Fe
2
O
3
recrystallization is further developed and the interconnection
between particles has been enhanced, and individual particles are scarce in pellet. The OH
particles have been connected with or even enclosed by SH particles, and the strength of
pellet is further increased. However, the inner structure of the OH particles are still
compact, with marked difference from the SH particles, which suggests that recrystallization
of Fe
2
O
3
in the OH particles is still undeveloped at 1230C, and the junctions between
particles are from the Fe
2
O
3
recrystallization of SH particles mainly.
At 1280C, it can be seen from Fig. 9 that Fe
2
O
3
recrystallization in the OH particles develops
satisfactorily; all of hematite particles become porous, the profile of the OH particles almost
disappears, and all the particles (both SH and OH) are connected with each other through
Fe
2
O
3
recrystallization and form a whole crystal structure; therefore, the strength of pellet is
further improved.
In summary, the results indicate that Fe
2
O
3
from SH and OH particles have different
activities during roasting, which results in the difference of strength formation mechanisms
of the mixed H/M concentrates pellets at various roasting temperatures. At lower
temperature (1150C), the pellet strength is mainly provided by Fe
2
O
3
recrystallization of the
SH particles. However, OH particles can be connected with the SH particles through crystal
bridges formed by high activity Fe
2
O
3
of SH particles, or even enclosed by the SH particles.
So, OH particles also contribute to the pellet strength to a certain degree.
When the temperature goes up to 1250C, Fe
2
O
3
recrystallization in the SH particles is
further developed, and the crystal junctions between the particles become stronger.
However, Fe
2
O
3
recrystallization within OH particles does not obviously take place, and few
joint is formed between the close OH particles. Pellet strength is mainly provided by Fe
2
O
3
recrystallization junctions of SH particles.
At 1280C, Fe
2
O
3
recrystallization in the SH and OH particles simultaneously occurs, and
the joints between the close particles, including OH and SH, are well developed. Particles
are connected with each other and the roasted pellet forms a whole crystal structure. The
Fe
2
O
3
recrystallization of SH and OH particles plays a crucial role in improving the roasted
pellet strength.
The results indicate that, because the activity of Fe
2
O
3
in the SH grains is higher than that in
the OH grains, Fe
2
O
3
recrystallization bonds between particles can be enhanced by the SH
grains during preheating and roasting, thus the newborn SH is able to improve the pellet
strength and decrease the roasting temperature of mixed H/M concentrates pellet. It is the
reason that adding a certain proportion of magnetite concentrate is for the enhanced
roasting performance of hematite pellet.
3. Fe
2
O
3
recrystallization during the firing of carbon-burdened hematite pellet
3.1 Materials and methods
3.1.1 Materials
The hematite concentrate is characterized by high total iron grade (67.2% TFe) and less
impurities (Table 3). The particle size is 92% undersize 0.074mm and the specific surface
area is 1629.5cm
2
/g (Table 4).
Recrystallization
340
Anthracite powder was used as the material of burdened carbon, of which the specific
surface area reaches 6599 cm
2
/g after grinding.
Materials Fe
total
FeO SiO
2
CaO MgO Al
2
O
3
LOI*
Hematite 67.22 0.55 2.17 0.01 0.05 0.55 0.59
Anthracite 1.48 / 6.86 1.51 0.24 5.88 84.08
Bentonite 7.07 / 60.61 0.94 2.2 17.98 10.41
*LOI-Loss on ignition
Table 3. Chemical compositions of the materials /%
+0.074mm/% 0.074-0.045mm/% -0.045mm/% specific surface area/cm
2
.g
-1
7.79 28.99 63.22 1629.5
Table 4. Size distribution and specific surface area of hematite concentrate
3.1.2 Methods
For each test, 5 kg of hematite concentrate was blended with the given proportion anthracite
powder, using 1.25% bentonite as binder. The green balls were prepared in a disc pelletizer
with a diameter of 1000 mm, and the green balls with 9~15 mm in diameter were statically
dried at 105C in a drying oven for 4 hours.
Firing teats were carried out in an electrically heated shaft furnace. To simulate firing
atmosphere, mixed gas of N
2
/O
2
at the given oxygen partial pressure (volume fraction) was
pumped into the shaft furnace at a certain flow-rate. The dry balls were charged into heat-
resistant pot, and then the pot was pushed downwards into the furnace step by step. The
pellets were fired at the given temperature for a given period. Afterwards, the roasted
pellets were taken out and naturally cooled to ambient temperature. The compression
strength of cooled pellets was measured with an LJ-1000 material experimental machine. An
average value of 10 pellets is expressed as the compression strength for each test.
3.2 Firing characteristics of carbon-burdened hematite pellet
3.2.1 Effects of anthracite dosage on compression strength
The effects of anthracite dosage on the compression strength of the roasted pellets are
shown in Fig. 10.
As shown in Fig. 10, the pellet strength with 0.5% anthracite is a little lower than that with
no anthracite. When anthracite dosage reaches 1.0%~1.25%, the compression strength goes
up to the maximum, however, the strength decreases greatly if the anthracite further
increases from 1.5% to 4%. The results indicate appropriate anthracite amount of
1.0%~1.25% may improve the strength of hematite pellet.
Recrystallization of Fe
2
O
3
During the Induration of Iron Ore Oxidation Pellets
341
0 1 2 3 4
1000
1500
2000
2500
3000
C
o
m
p
r
e
s
s
i
o
n
s
t
r
e
n
g
t
h
o
f
p
e
l
l
e
t
(
N
/
P
)
Anthracite dosage (%)
(Oxygen partial pressure: 20%, airflow: 6L/min, roasting temperature: 1280C, roasting time: 20min)
Fig. 10. Effects of anthracite dosage on the compression strength of pellet
3.2.2 Effects of roasting temperature on compression strength
The effects of roasting temperature on compression strength of pellet are shown in Fig. 11.
1000 1100 1200 1300 1400
0
500
1000
1500
2000
2500
3000
3500
C
o
m
p
r
e
s
s
i
o
n
s
t
r
e
n
g
t
h
o
f
p
e
l
l
e
t
(
N
/
P
)
Roasting temperature ()
With 1% anthracite
Without anthracite
(Oxygen partial pressure 20%, airflow: 6L/min, roasting time: 20min)
Fig. 11. Effects of roasting temperature on the compression strength of pellet
Recrystallization
342
As shown in Fig.11, the strength of carbon-burdened pellet is always higher than that of
pellet without carbon when roasted at the same temperature; moreover, for carbon-
burdened hematite pellet, the compression strength increases constantly with the
temperature. However, the strength of pellet in the absence of carbon not only doesnt
increase markedly until 1250C, but also decreases over 1320C due to the decomposition of
Fe
2
O
3
. The results imply that the roasting temperature can be decreased and the firing
temperature range is enlarged by adding an appropriate amount of anthracite into hematite
pellet.
3.2.3 Effects of roasting time on compression strength
The compression strength of roasted pellet increases gradually with the roasting time
prolonging and a maximum strength is obtained at 25 min (in Fig.12), hereafter, the strength
almost remains constant.
15 20 25 30
2000
2400
2800
3200
C
o
m
p
r
e
s
s
i
o
n
s
t
r
e
n
g
t
h
o
f
p
e
l
l
e
t
(
N
/
P
)
Roasting time (min)
(Oxygen partial pressure: 20%, airflow: 6L/min, roasting temperature: 1280C, anthracite: 1.0%)
Fig. 12. Effects of roasting time on the compression strength of carbon-burdened pellet
3.2.4 Effects of oxygen partial pressure on compression strength
The compression strength of pellet is susceptible to the change of oxygen partial pressure as
shown in Fig.13. The strength of the pellet roasted in N
2
, that is, in the inert atmosphere, is
slightly higher than that in 10% O
2
. The strength nearly reaches the maximum at 20% O
2
,
and then decreases gradually with increasing the oxygen partial pressure.
Recrystallization of Fe
2
O
3
During the Induration of Iron Ore Oxidation Pellets
343
0 10 20 30 40 50
1500
2000
2500
3000
C
o
m
p
r
e
s
s
i
o
n
s
t
r
e
n
g
t
h
o
f
p
e
l
l
e
t
(
N
/
P
)
Oxygen partial pressure (%)
(Airflow: 6L/min, roasting temperature: 1280C, roasting time: 20min, anthracite: 1.0%)
Fig. 13. Effects of oxygen partial pressure on the compression strength of carbon-burdened
pellet
3.3 Roles of burdened carbon during the roasting
To shed light on the effects of the burdened carbon on the induration of hematite pellet, a
test, as shown in Fig. 14, was designed to identify the reduction and decomposition of
hematite in carbon-burdened pellet during roasting.
As shown in Fig. 14, a cylinder was made by briquetting hematite concentrate firstly, the
cylinder bottom is closed, and its inner diameter is 20 mm, the outer diameter is 30 mm. To
allow the gas upward injecting into the inner cylinder and penetrating through the
anthracite powder layer, many ventages with 0.1 mm diameter were drilled through the
cylinder bottom.
Dry pellets with 2~3 mm in diameter were prepared from hematite concentrate in advance,
and then were charged into the surface layer of inner cylinder. The cylinder bottom,
anthracite powder layer and pellet layer were separated by inert material of Al
2
O
3
powder
to avoid their contact with each other.
At the beginning of trail, the sample prepared according to Fig. 14 was placed in an
electrically heated shaft furnace, and 6L/min N
2
with 99.99% purity was pumped into the
shaft furnace from the bottom. The sample was taken out and immersed into water
immediately after roasted at 1280C for 20 minutes. Subsequently, the FeO content of the
cylinder bottom and pellet was measured respectively.
It is shown that FeO content of the cylinder bottom and pellet is 4.35% and 28.69%
respectively. FeO content of the pellet is obviously higher than that of the cylinder bottom.
Recrystallization
344
Fig. 14. Schematic diagram of the test on the role of burdened carbon during roasting
In N
2
atmosphere, hematite may decompose into magnetite and release O
2
according to formula
(1), and FeO content increases accordingly. It is the reaction that the hematite within the
cylinder bottom occurs. However, because of being separated by Al
2
O
3
powder, the hematite
within the cylinder bottom cant be reduced by anthracite or upwards flowing reductive gases
CO/H
2
, which are produced by gasification of anthracite. Therefore, the increase of FeO content
in the cylinder bottom is only caused by the decomposition of hematite.
However, as regards as the hematite within the pellet, on one hand, it may be decomposed
into magnetite in N
2
gas as same as the hematite within the cylinder bottom; on the other
hand, it can be also reduced into magnetite by upwards flowing CO/H
2
produced by
gasification of anthracite. Therefore, the increase of FeO content of the pellet is caused by
both the decomposition and the reduction of hematite, and the latter is more crucial.
The above results identify that the burdened carbon plays the role of reductant during the
roasting, and a large number of newborn magnetite are created due to the reductive reaction
of hematite by CO/H
2
, the products of gasification of anthracite.
Of cause, the burdened carbon can release heat via combustion and heat up the inner pellet,
which is advantageous for the induration of hematite pellet. The heating function of the
burdened carbon resembles the heat release by oxidation of the magnetite concentrate
added into hematite pellet.
Recrystallization of Fe
2
O
3
During the Induration of Iron Ore Oxidation Pellets
345
3.4 Induration mechanisms of carbon-burdened hematite pellet
3.4.1 Changes of FeO content during the roasting of carbon-burdened pellet
3.4.1.1 Effects of anthracite dosage
FeO content of the roasted pellets with different dosage of anthracite is measured by
chemical analysis and the results are shown in Fig. 15.
As shown in Fig.15, the FeO content of the roasted pellets increases gradually with the
increasing the dosage of anthracite. This suggests that the anthracite in pellet benefits the
reduction of hematite during roasting. Because of the relatively strong reductive atmosphere
under the condition of high dosage of anthracite, a large amount of Fe
2
O
3
is reduced into
Fe
3
O
4
.
The above results that FeO content of the pellet varies with the anthracite dosage are in
accord with the conclusions obtained from the tests as shown in Fig. 14
0 1 2 3 4
0
6
12
18
24
F
e
O
c
o
n
t
e
n
t
o
f
p
e
l
l
e
t
(
%
)
Anthracite dosage (%)
(Oxygen partial pressure: 20%, airflow: 6L/min, roasting temperature: 1280C, roasting time: 6min)
Fig. 15. Effects of anthracite dosage on FeO content of the carbon-burdened pellet
3.4.1.2 Effects of oxygen partial pressure
FeO content of the pellets roasted at different oxygen contents is measured and the results
are shown in Fig. 16.
As shown in Fig.16, the reduction of hematite mainly occurs at the initial roasting stage. The
time, when the maximum FeO content attains to, is shortened with increasing the oxygen
partial pressure, which indicates that the increase of oxygen partial pressure has favourable
effect on oxidization rate of newborn magnetite.
Recrystallization
346
0 5 10 15 20 25
0
5
10
15
20
25
O
2
content
0%(N
2
)
5%
10%
20%
30%
40%
50%
F
e
O
c
o
n
t
e
n
t
o
f
p
e
l
l
e
t
(
%
)
Roasting time (min)
(Anthracite: 1.0%, airflow: 6L/min, roasting temperature: 1280C)
Fig. 16. Change of FeO content of the pellets roasted at different oxygen partial pressures for
different period
The maximum FeO content decreases when the carbon-burdened hematite pellet is
roasted at high oxygen partial pressure. The result shows that the lower the oxygen
partial pressure during roasting, the stronger the reductive atmosphere within the pellet;
and the reduction of Fe
2
O
3
can be enhanced. At high oxygen partial pressure, the
burdened carbon in pellet combusts completely, and more carbon comes into being not
CO but CO
2
, thus reductive atmosphere is weakened, which prevents the hematite from
reducing into magnetite.
In oxidative atmosphere, FeO content of the pellet increases firstly and then decreases
during roasting, the reason for which is that, at the initial roasting stage, the reduction rate
of hematite into magnetite is higher than the oxidation rate of newborn magnetite into
hematite, however, accompanied with the consumption of burdened carbon, the reduction
rate decreases while the oxidation rate increases, the maximum FeO content attains when
the reduction rate equals to the oxidation rate. Subsequently, FeO content decreases alone
with the oxidation of magnetite untill the oxidation is complete.
It indicates that some original hematite (OH) can be reduced to magnetite by the burdened
carbon firstly, and then the newborn magnetite is oxidized into secondary hematite (SH)
again during the roasting. The transformation of OH to magnetite and newborn magnetite
to SH changes the route of Fe
2
O
3
recrystallization during the induration of carbon-burdened
hematite pellet.
Recrystallization of Fe
2
O
3
During the Induration of Iron Ore Oxidation Pellets
347
3.4.2 Fe
2
O
3
recrystallization behaviours of carbon-burdened hematite pellet
To shed light on the Fe
2
O
3
recrystallization behaviour, the microstructure of carbon-
burdened hematite pellets at different roasting stage is investigated and the results are
presented in Fig. 17.
Recrystallization
348
Roasting time: a-4min; b-6min; c-8min; d-20min
1-Magnetite; 2-OH3- SH
(Anthracite: 1.0%, Oxygen partial pressure: 20%, airflow: 6L/min)
Fig. 17. Microstructure of the carbon-burdened pellets roasted at 1280C
Recrystallization of Fe
2
O
3
During the Induration of Iron Ore Oxidation Pellets
349
When the volatile matter in anthracite is pyrolyzed and gasified during initial roasting
stage, the reduction atmosphere is gradually enhanced within the pellet, and a few
magnetite grains, shown in Fig. 17a, are formed by the reduction of OH. An apparent colour
difference can be observed between the hematite grains and magnetite grains. The hematite
presents bright and white colour, while the magnetite is grey. However, the OH particles
keep their original shape and discernible angularity. The inner crystal structure of OH is
compact and there is no crystal bond formed between OH particles.
It can be seen from Fig. 17b, a large amount of newly created magnetite is observed as the
roasting process progresses; however, the OH particles keep their original shapes, and there
is no crystal junction observed still.
As shown in Fig. 17c, the content of newborn magnetite decreases, and a few SH, which
comes from the oxidation of the newborn magnetite, are formed. In comparison with OH
grains, the angularities of the SH particles become unclear, and a few crystal junctions can
also be observed between them. In this stage, with the consumption of anthracite, reduction
atmosphere is weakened and oxidation rate of SH is accelerated.
At the end of roasting, the newborn magnetite has been completely oxidized into SH, and a
large number of crystal junctions between particles are formed (as shown in Fig. 17d). Pellet
strength is mainly provided by Fe
2
O
3
recrystallization junctions between SH particles and
OH particles.
It can be concluded from the results mentioned above that partial OH particles can be
reduced firstly and turn into magnetite particles by the anthracite powder dispersed in the
carbon-burdened hematite pellet, however, the newborn magnetite can be subsequently
oxidized into SH particles with higher activity, and the route of Fe
2
O
3
recrystallization is
changed from the recrystallization of OH particles to the recrystallization bonds among SH
particles and OH particles. Therefore, the formation of SH during the roasting of carbon-
burdened hematite pellet is able to improve the pellet strength and decrease the roasting
temperature. It is the reason why adding a certain proportion anthracite is also an effective
way to enhance the roasting performance of hematite pellet.
4. Conclusions
Both magnetite and burdened carbon are found to be the favourable techniques for
enhancing the induration of hematite pellet. The induration mechanisms of hematite pellet
with addition of magnetite concentrate and anthracite powder are revealed by
characterization of Fe
2
O
3
recrystallization rules during the oxidization roasting.
The crystallization behaviours of Fe
2
O
3
during preheating and roasting of pellets made from
mixed hematite/magnetite (H/M) concentrates have been revealed. The results indicate that
the strength of pellet is mainly provided by the crystalline connections between the particles
during preheating. This occurs because the activity of Fe
2
O
3
from secondary hematite
(oxidized from magnetite concentrate) is higher than that from original hematite (from the
raw hematite concentrate). In the roasting process, when temperature is lower than 1250C,
the strength is mainly provided by the development, connection and growth of Fe
2
O
3
crystalline grains from secondary hematite. Only if the temperature exceeds 1280C does
Fe
2
O
3
recrystallization in original hematite grains develop very well.
Recrystallization
350
Because the activity of Fe
2
O
3
in the secondary hematite grains is higher than that in the
original hematite grains, Fe
2
O
3
recrystallization bonds between particles can be enhanced by
the secondary hematite grains during preheating and roasting, and the secondary hematite
in H/M concentrate pellet is able to improve the strength and decrease the roasting
temperature of hematite pellet. Thus, adding a certain proportion of magnetite concentrate
is an effective way to improve the roasting performance of hematite pellet.
The effects of anthracite on oxidation roasting behaviour for hematite pellet have been
elucidated. Anthracite in pellet has two functions: the one lies on that a part of heat needed
in roasting process can be supplied by the carbon combustion, and the other is that the
reduction of partial hematite by the carbon dispersed in pellet, as well as the partial
decomposition of hematite at relatively low oxygen partial pressure, leads to the
transformation of hematite into magnetite during the roasting.
Based on microstructure analysis, it can be founded that the new-born magnetite, produced
from the reduction and the decomposition of original hematite (OH), is oxidized into the
secondary hematite (SH) during roasting. Thus, Fe
2
O
3
recrystallization bonds between
particles can be consolidated by the secondary hematite grains at lower roasting temperature.
The strength of carbon-burdened hematite pellet is enhanced and the roasting temperature is
decreased due to the formation of secondary hematite. Therefore, adding a certain proportion
of anthracite is also an effective way to enhance the roasting performance of hematite pellet.
5. Acknowledgments
The authors want to express their thanks to National Science Fund for Distinguished Young
Scholars (50725416), National Natural Science Foundation of China (50604015 and 50804059)
and Fundamental Research Funds for the Central Universities for financial supports of this
research. Dr. Mingjun Rao is appreciated for his helpful remarks on spelling and expression.
6. References
Jiang, T.; Zhang, Y. & Huang, Z. (2008). Preheating and Roasting Characteristics of
HematiteMagnetite (HM) Concentrate Pellets. Ironmaking Steelmaking, Vol. 35,
No. 1, pp. (21-26), ISSN 0301-9233
Li, G.; Li, X. & Zhang, Y. (2009). Induration Mechanisms of Oxidised Pellets Prepared from
Mixed MagnetiteHaematite Concentrates. Ironmaking Steelmaking, Vol. 36, No. 5,
pp. (393-396), ISSN 0301-9233
Xu, M. (2001). Deve1opment of BF Burden Structure in China. Ironmaking, Vol. 20, No. 2, pp.
(24-27), ISSN 1001-1471
Ball, D.; Dartnell, J. & Davison, J. (1973). Agglomeration of Iron Ores, Heinemann Educational,
ISBN 0435720104, London
APbill, J. (1985). Carbonaceous Additives for Pelletizing Production, 4
TH
International
Symposium on Agglomeration, ISBN 0-932897-00-2, Toronto, Canada, June, 1985
Clout, J.; Manuel J. (2003). Fundamental Investigations of Differences in Bonding
Mechanisms in Iron Ore Sinter Formed from Magnetite Concentrates and Hematite
Ores. Powder Technology, Vol. 130, pp. (393-399), ISSN 0921-8831
Yang, X.; Guo, Z. & Wang, D. (1995). Research on the Reduction Mechanism of Iron Ore
Pellets Containing Graphite. Engineering Chemistry and Metallurgy, Vol. 16, No. 2,
pp. (118-126), ISSN 1001-2052
14
Ion-Beam-Induced Epitaxial Recrystallization
Method and Its Recent Applications
Rossano Lang
1,4
, Alan de Menezes
1
, Adenilson dos Santos
2
,
Shay Reboh
3,4
, Eliermes Meneses
1
, Livio Amaral
4
and Lisandro Cardoso
1
1
Instituto de Fsica Gleb Wataghin - UNICAMP, Campinas, SP
2
CCSST, Universidade Federal do Maranho, Imperatriz, MA
3
Groupe nMat, CEMES-CNRS, Toulouse
4
Instituto de Fsica - UFRGS, Porto Alegre, RS
1,2,4
Brazil
3
France
1. Introduction
The transition from amorphous Silicon to crystalline Silicon is a process of great
technological importance and has raised an enormous interest also from a purely scientific
perspective. Ion irradiation through an amorphous/crystalline interface may stimulate
recrystallization or layer-by-layer amorphization depending on the sample temperature and
ion beam parameters. In this chapter, we address some key features of this recrystallization
phenomenon. The recrystallization/amorphization process will be discussed in relation to
its dependence on the energy deposited during the ion beam irradiation, the sample
temperature, and the presence of impurities, such as iron atoms dissolved within an
amorphous silicon layer. Also it will be discussed the specific experimental condition under
which a metastable phase of the FeSi
2
binary compound is trapped within a region of the
recrystallized Silicon, in the form of nanoparticles. Theses nanoparticles, with different
orientations and morphologies, are shown to cause interesting distortions in the
surrounding Si crystal lattice. This effect reflects an important application of the ion-beam
induced recrystallization process, as a method that could be used to synthesize ordered
nanoparticles within a Silicon matrix.
2. Epitaxial crystallization
The amorphous Silicon (a-Si) is a Si phase with well-defined thermodynamic properties
and that presents Gibbs free energy of ~ 0.12 eV/at. higher than that of the crystalline
phase (c-Si) (Donovan et al., 1985, 1989; Roorda et al., 1989). This implies the existence of a
driving force for the transition from the amorphous to the crystalline phase to occur. In
other words, there is a natural tendency for the (a-Si) (c-Si) transformation. At room
temperature, the a-Si phase is metastable and it is transformed into c-Si only when
submitted to high temperatures, typically higher than 450 C (Olson & Roth, 1988;
Williams, 1983). For the case of an amorphized layer on top of a Si single-crystal substrate,
Recrystallization
352
the transition occurs by a planar motion of the crystal-amorphous (c-a) interface, from the
interior towards the surface, and hence decreasing the thickness of the existing
amorphous layer with time, as schematically illustrated in Fig. 1. This effect is called
epitaxial crystallization. For a pure thermal recrystallization process - SPEG (Solid Phase
Epitaxial Growth), the growth rate is strongly dependent on the temperature and it
presents an Arrhenius-like behavior with a unique activation energy of (2.68 0.05) eV
over a growth rate range of more than six orders of magnitude (Olson & Roth, 1988). For
instance, at 470 C the c-a interface displacement velocity is ~ 1 /min while at lower
temperatures, this value decreases considerably and the amorphous to crystal transition
becomes kinetically inhibited. However, epitaxial recrystallization of a-Si layers can also
be achieved at lower temperatures (~ 200 - 320 C) by ion-beam irradiation. This
phenomenon represents a typical example of a dynamic annealing process and it is
usually referred to as Ion-Beam-Induced Epitaxial Crystallization - IBIEC (Priolo & Rimini,
1990).
It is well known that both the implantation and the irradiation processes (depending on the
dose and the beam ion mass) have the amorphization of the host matrix, as one of their main
effects. For the specific case of Si, when the deposited energy (energy loss of ion beam
mainly by nuclear collisions) handed by the projectile exceeds the threshold value of ~ 13
eV/at. a buried amorphous Si layer is formed (Narayan et al., 1984). However, if the Si is
irradiated at a temperature above ~ 150 C, there is a competition between both phenomena:
amorphization and recrystallization.
Fig. 1. Schematic illustration of two recrystallization processes in Silicon. SPEG (Solid Phase
Epitaxial Growth) is a purely thermal process while, IBIEC is an ion beam induced epitaxial
recrystallization process.
In fact, the temperature rise of the Si substrate is not solely responsible for this trend of the
crystalline order recovery, since the recrystallization rate is greater than that obtained by a
purely thermal process at the same temperature. A more complex mechanism involving a
dynamic reordering stimulated by the ion beam is responsible for this effect. The
recrystallization under ion-beam irradiation is the result of the prevalence of dynamic
annealing over defect generation rate. Both processes depend on the substrate temperature
and on the parameters of the irradiating beam, such as ion species, energy, ion dose and ion
flux - also known as dose rate.
Ion-Beam-Induced Epitaxial Recrystallization Method and Its Recent Applications
353
2.1 Crystal-amorphous interface displacement velocity
Rutherford backscattering spectrometry in combination with the ion channeling technique
(RBS/C) is commonly used to directly monitor the c-a interface motion and infer the kinetics
of the IBIEC process. The following RBS/C results were carried out with He
+
beam at 1 MeV
in a 170 scattering geometry. Figure 2 presents c-a interface position measurements of [100]
oriented Si substrate previously implanted at room temperature with Fe
+
ions at 100 keV
energy. Channeled implantation was avoided by tilting the sample 7 normal with respect to
the incident beam direction. Subsequently, the recrystallization of the Fe-implanted surface
amorphous Si layer (filled circles in the figure) was induced by Si
+
irradiation at high energy
- 600 keV (whose projected range is well beyond the original c-a interface). In the IBIEC
experiments the substrate temperature was fixed and controlled. This allows discrimination
between the effects due to the heating of the sample holder and those due to ion-beam
irradiation. In order to avoid beam heating effects low current density (~ 1 A/cm
2
) was
therefore used with ion flux of ~ 6.2x10
12
ions/cm
2
s. The substrate temperature was
maintained at 350 C. Essentially, one observes that the increase of the Si
+
irradiated dose,
leads to a decrease in the distance between the c-a interface and the surface. That is, there is
a recrystallization towards the surface.
Fig. 2. Determination of the thickness of an amorphous layer using the ion channeling
technique. This particular case corresponds to the recrystallization of ~ 176 nm a-Si under
600 keV Si
+
beam irradiation keeping the substrate at 350 C.
In implantation processes (Dearnaley et al., 1973), the implanted dose is the integral of the
ion flux in time, which corresponds to the total number of ions that were focused on the
sample per unit area
2
2
0
(ion/cm ) ; (s)
A/cm
T
e
e
nq I
dt T
nq
= =
}
, (1)
Recrystallization
354
where: I is the beam current in ampere per unit area of the sample (A/cm
2
), q
e
is the electron
charge, n = 1 for once ionized ions, n = 2 for doubly ionized species, and so on. The term I/q
e
is designated as the ion flux | (ion/cm
2
s) and T(s) is the implantation time in seconds. For
the IBIEC, one designates a certain irradiation dose and, consequently, the ion flux for
such irradiation by | = d/dt. When the ion flux is kept constant at a given crystallization
procedure, the irradiation dose may be interpreted as a measure of the processing time
described by
d
d
c a c a c a
X dX dX d
t d dt d
|
= = . (2)
Several reports define a c-a interface velocity R as the derivative of the recrystallized
thickness (nm) versus dose (at/cm
2
) curve. Thus, the unit of such "velocity" is expressed in
nm/(at/cm
2
). In the present study, figure 3 is an example that enables us to extract the
interface displacement velocity or recrystallization rate.
Fig. 3. Regrown thickness for [100] Si amorphized by Fe
+
ions and subsequently irradiated at
350 C by 600 keV Si
+
ions. The interfacial displacement velocity R can be extracted from the
curve.
It should be noted that a comparison between the thermal crystallization rate V (dX
c-a
/dt)
and the recrystallization rate defined for the IBIEC R (dX
c-a
/d) is always possible, if one
knows the ion flux | of the process as shown in equation (2).
2.2 Influence of the beam parameters in the IBIEC process
Several experimental results have indicated that the beam parameters such as ion species
(by the factor of nuclear energy loss S
n
) and the ion flux | have direct influence on the
recrystallization rate. With regards to the energy loss by projectile-target elastic collisions,
the following result is quite interesting. By using the recrystallized thickness versus
irradiation dose plot (Fig. 3), the recrystallization rate R as a function of depth was
Ion-Beam-Induced Epitaxial Recrystallization Method and Its Recent Applications
355
determined as shown in figure 4. In the same figure, one observes the number of vacancies
produced by irradiation of Si
+
ions as a function of depth. This evaluation corresponds to a
calculation obtained by the SRIM algorithm with a displacement energy of 15 eV and a
lattice binding energy of 2 eV) (Ziegler, 2011). The depth dependence of the regrowth
experimental rates seems to follow the profile of defect generation.
Fig. 4. Regrowth rate versus depth for a Si layer amorphized by Fe
+
ions and recrystallized
at 350 C by 600 keV Si
+
ions. The inset shows the number of vacancies generated by Si
irradiation calculated by the SRIM code (closed circles).
This result suggests that the ion-induced recrystallization rate is associated with the
production of point defects, or in a more general way, with the energy loss into elastic
collisions at the c-a interface. Therefore, only those defects generated nearby or directly at
the c-a interface are available for the recrystallization process. However, it should be
mentioned that the diffusion or vacancy migration is not the promoter mechanism of the
IBIEC. This issue will be addressed later. The abovementioned conclusions were also
confirmed by other experimental results (Holmn et al., 1984; Linnros et al., 1984, 1985;
Miyao et al., 1986; Williams et al., 1985). For instance, Linnros (Linnros et al., 1985) reported
a clear experimental evidence of a linear dependence of the ion-induced recrystallization
rate on the nuclear energy loss. In their experiments, ion beams of different masses (He, N,
Ne, Si, As, Kr) were used to stimulate IBIEC where the recrystallization rate was observed to
increase with increasing ion mass. This demonstrates that nuclear energy loss is the
mechanism which produces the defects responsible for the IBIEC. Furthermore, the effects
produced by both kinds of energy loss (nuclear and electronic) were discriminated through
the dependence of the recrystallization rate on the beam energy. It was inferred that the
electronic excitations and ionizations play practically no role in the recrystallization process
(Elliman et al., 1985; Williams et al., 1985). In other experiments, in which a-Si layers were
irradiated by the electron-beam, the recrystallization was observed only for energies above a
threshold for atomic displacement of ~ 150 keV while below this threshold no epitaxial
Recrystallization
356
regrowth was produced, even after irradiation at very high electron doses (Lulli et al., 1987;
Miyao et al., 1986; Washburn et al., 1983). Therefore, the observed epitaxy is associated with
elastic collisions that transfer sufficient momentum to displace target atoms from their
lattice site.
As mentioned earlier in this section, there is another parameter that directly influences the
process - the ion flux. In general, it is observed that the lower the flux |, the higher is the
effective velocity R (dX
c-a
/d) defined by equation 2. This parameter is so important that
high fluxes almost inhibit the process, especially for heavier ions (Linnros et al., 1985).
2.3 Temperature dependence
The sample temperature during irradiation is a fundamental variable in the IBIEC process
(Elliman et al., 1985; Linnros et al., 1984; Priolo et al., 1988, 1989; Williams et al., 1985).
Figure 5 shows, in an Arrhenius plot, the ion-induced growth rate (or recrystallization) as a
function of the reciprocal temperature of an a-Si/Si(100) layer recrystallized by 600 keV Kr
2+
ion irradiation (dose: 1x10
15
Kr/cm
2
and ion flux: 1x10
12
Kr/cm
2
s). It also shows the
recrystallization rate that represents the thermal contribution (SPEG) with an activation
energy of (2.68 0.05) eV. The data were extracted from references (Olson & Roth, 1988;
Priolo et al., 1988) and reported in the figure. The growth rate is reported in /s (left-hand
side) and in
4
/eV (right-hand side). The latter scale represents the growth rate in the form
X/(E), X being the recrystallized thickness, the dose and (E) the total energy
deposited responsible for the displacement production at the c-a interface.
Fig. 5. Ion-induced growth rate versus reciprocal temperature for a-Si layers produced by Si
+
implantation and recrystallized by 600 keV Kr
2+
ion irradiation. The thermal contribution to
the growth rate is also displayed. Data extracted from references (Olson & Roth, 1988; Priolo
et al., 1988) and reported in the figure.
Ion-Beam-Induced Epitaxial Recrystallization Method and Its Recent Applications
357
It should be noted that recrystallization due to ion-beam irradiation occurs in a temperature
range for which the thermal process is kinetically inhibited. Therefore, ion-beam irradiation
strongly enhances the kinetics of recrystallization. For instance, at 250 C the ion-induced
growth rate is 0.07 /s while, an extrapolation of the thermal data gives a rate of only 10
-10
/s. Furthermore, in the temperature range shown, the IBIEC presents an Arrhenius-like
temperature dependence with an apparent activation energy of (0.32 0.05) eV as also
demonstrated by several other experiments (Elliman et al., 1985; Linnros et al., 1984;
Williams et al., 1985).
It has been proposed that the activation energy for thermal recrystallization of a-Si layers is
composed of two terms: one for defect generation and other for defect migration. Ion-beam
irradiation clearly removes the main activated process usually limiting conventional thermal
regrowth. In fact, during ion-beam irradiation defects are not thermally generated but rather
being produced by means of atomic collisions. Jackson (Jackson, 1988) proposed that the
activation energy inferred from the IBIEC is not associated with any activated process and
therefore considered just as an apparent activation energy. The linear dependence of the
regrowth rate as a function of reciprocal temperature, in a logarithmic plot, comes instead,
from a balance between different effects. In addition, Jackson (Jackson, 1988) in his
intracascade model suggested the dangling bond in the amorphous phase as the promoter
for IBIEC. Dangling bonds are structural defects which by moving nearby the c-a interfacial
region should produce a rearrangement of the bonds leading to recrystallization. In the mid-
1990s, Priolo (Priolo et al., 1990) proposed a phenomenological model of IBIEC which
combines the approach of the Jackson model (Jackson, 1988) with the structural and
electronic features of models proposed for conventional thermal regrowth (Williams &
Elliman, 1986). This model has explained all the experimental results so far.
2.4 Planar amorphization
In the previous sections, it was shown how a combination of thermal energy and energy
deposited by ballistic effects can produce a non-equilibrium epitaxial recrystallization.
However, IBIEC is a reversible process, where the increase in the ion flux of the irradiating
beam, and /or the decrease in the target temperature can cause a planar layer-by-layer
amorphization instead of an epitaxial recrystallization (Elliman et al., 1987; Linnros et al.,
1986, 1988). Both processes are schematically illustrated in figure 6. At a constant flux, there
is a substrate temperature T
R
(reversal temperature) such that, when T > T
R
, the ion
irradiation produces epitaxial regrowth whereas, when T < T
R
the irradiation produces a
layer-by-layer amorphization. The remarkable point is that the amorphization occurs just
from the pre-existing amorphous seed and not from regions below the c-a interface, despite
the fact that the energy loss by elastic collisions increases with the increasing depth. In this
context, the net velocity of the interface motion can be described by two terms: a
crystallization and an amorphization, as
d d d
/ /
d d d
c a c a c a
dX X X X
R
d t t t
| |
(
= = =
(
. (3)
When R is positive, the crystallization regime is prevalent and when R is negative the
system is in the amorphization regime.
Recrystallization
358
Fig. 6. Schematic representation of the ion-induced crystallization and amorphization. At
temperatures below T
R
the amorphous layer grows (amorphization regime), whereas at
temperatures above T
R
, it shrinks (crystallization regime).
An example of planar amorphization is shown in figure 7. In this example, a ~ 90 nm thick
layer onto a Si(100) substrate was amorphized at room temperature by 40 keV Fe
+
ions at a
dose of 1x10
16
/cm
2
. The irradiation was performed by using a 380 KeV Ne
+
beam at a dose
of 1x10
17
/cm
2
and ion flux of ~ 1.5x10
13
/cm
2
s. The substrate temperature was fixed at 100
C. In the figure two RBS spectra are displayed in channeling condition for the a-Si layer
before (open circles) and after (closed circles) Ne irradiation. A random spectrum is also
reported. Ne irradiation clearly produces a great amount of damage beyond the original c-a
interface. The a-Si surface layer is clearly seen to enlarge under a planar motion towards the
sample interior. The amorphous layer has become ~ 50 nm thicker.
2.5 Impurity effects
Besides temperature and beam parameters, the kind of impurity dissolved in the sample
plays an important role in the IBIEC. The presence of impurities dispersed within the a-Si
layer can dramatically affect the recrystallization process. Depending on their behavior, they
can be divided in two major categories: fast and slow diffusers. Fast diffusers comprehend
species like Cu, Ag and Au which, at typical temperatures of the IBIEC process (~ 300 C)
have diffusivities of the order of 10
-12
- 10
-15
cm
2
/s and low solid solubility. These impurities
have therefore enough mobility to be redistributed at the advancing c-a interface during
recrystallization, modifying the impurity initial profile through the segregation towards the
surface imposed by the planar advance. On the other hand, the slow diffusers such as B, P,
As, do not present the effect of segregation - are immobile in the time-temperature windows
Ion-Beam-Induced Epitaxial Recrystallization Method and Its Recent Applications
359
used during the IBIEC experiments, since the interface displacement velocity is much higher
than the values of their mobilities in amorphous silicon. In this case, the initial concentration
profile of these impurities remains unchanged after recrystallization. This allows one to
produce non-equilibrium structures with impurities trapped in the c-Si at concentrations
well above their solid solubility. However, the presence of impurities at the c-a interface can
modify the recrystallization rate which is observed to increase or decrease, according to the
particular species and to its concentration.
Fig. 7. RBS channeling spectra along the [100] Si axis for a ~ 90 nm thick amorphous layer
before (open circles) and after (closed circles) 380 keV Ne
+
irradiation at 100 C and dose of
1x10
17
/cm
2
. Note the planar growth of the amorphous phase.
From this brief discussion, it is noticed that the behavior of crucial parameters in the IBIEC
process such as temperature and impurities is very similar to that in solid-phase epitaxial
growth case. Despite the fact that the impact of these parameters is quite different for each
phenomenon, there is a strong indication that similar microscopic processes occur in both
cases. More precisely, the same interfacial defect responsible for the thermal recrystallization
is considered to be active in IBIEC, being the ion beam the precursor of the increase in the
average concentration of these defects.
3. Nanoparticles synthesis by IBIEC
The IBIEC technique has been used as a method to synthesize nanoparticles in Silicon matrix
[Lang et al., 2010a, 2010b]. Specifically, the structural properties of the FeSi
2
nanoparticles
synthesized in Fe
+
low dose implanted Si(100) substrates were investigated. Particularly in
this experimental observation, the Fe proved to be a fast diffuser for IBIEC, despite of its
action as a retardant of the process, whose recrystallization rate was dependent on the
implanted Fe
+
concentration. Nevertheless, only the main results are reported here. The
remarkable results which are presented show that the shape of the synthesized material
Recrystallization
360
(observed by high resolution transmission electron microscopy - HRTEM) singularly affects
the surrounding Si lattice. The lattice strain shape-dependent distribution in both directions:
out-of-plane and in-plane was tailored by X-ray Bragg-Surface Diffraction technique.
3.1 Fe
+
ion implantation in Si(100) and recrystallization process
A Si(100) n-type Czochralski wafer (thickness 500 m, resistivity 10 - 20 cm) was used as
host matrix. Fe
+
ions at 40 keV were implanted at room temperature at an ion dose of 5x10
15
cm
-2
. Channeling effects were avoided by tilting the sample 7 normal with respect to the
incident beam direction. The typical iron beam current density during implantation was
about 150 nA/cm
2
. Subsequently, ion-beam recrystallization experiments were performed at
350 C using a 600 keV Si
+
beam (current density ~ 1 A/cm
2
) to a total dose of 6x10
16
ions/cm
2
. The dose rate resulting from the Si beam current was 6.2x10
12
ions/cm
2
s. It is
worth noting that Si
+
ions at 600 keV energy have a projected range (average depth) of ~ 770
nm with a straggle (standard deviation) of ~ 150 nm. These values ensure that the
irradiation exceeds the pre-existing amorphous layer (well beyond the original c-a
interface).
The structures obtained into the as-implanted and recrystallized samples were analyzed and
characterized by Rutherford Backscattering Spectrometry combined with ion channeling
technique (RBS/C - with He
+
beam at 1 MeV in a 170 scattering geometry) and also by
transmission electron microscopy (TEM - JEOL 2010 operating at 200 kV). High-resolution
rocking curves, as well as the reflection mappings of the Bragg-Surface Diffraction
reflections, i.e., at the exact multiple diffraction condition, were carried out using the Huber
multiaxis diffractometer of the XRD1 beam-line (Brazilian Synchrotron Radiation Facility -
LNLS), with an incident beam wavelength of = 1.5495(5) , as defined by using a Si(111)
channel-cut monochromator.
3.1.1 RBS and TEM Results and discussion
Figure 8 shows RBS/C spectra obtained at random and [100]-channeled direction from the
samples before (as-implanted) and after (recrystallized) irradiation. As observed in the
aligned as-implanted spectrum, the implantation has produced an amorphous layer over ~
90 nm while no channeling in the Fe signal was observed. However, expressive reduction of
dechanneling yield in the Si profile is detected after IBIEC process (aligned IBIEC spectra
relative to aligned as-implanted spectra). This decrease reflects the crystalline order
recovery. The entire amorphous Si layer was recrystallized (at an average rate of ~ 0.04
nm/s) with a minimum backscattering yield on the subsurface region being
min
6.3 %.
The RBS concentration-depth profile (not shown here) has indicated an implanted Fe peak
concentration of ~ 2.8 at.% at 40 nm from the surface. However, the recrystallization process
caused a slight narrowing of the Fe peak and a small segregation towards the surface. A
significant degree of channeling (
Fe
46 %) was also observed in the Fe spectrum.
Figure 9 shows bright-field TEM cross-section images (taken at [110]
Si
zone axis) of the as-
implanted sample. One clearly observes (Fig. 9a) the amorphous Si layer of ~ 90 nm
produced by 40 keV Fe
+
implantation and the crystal-amorphous interface. Figure 9b shows
in detail the c-a interface, where one can note a large amount of defects called "end-of-range
defects" generated by the implantation process. This defective intermediate zone between
Ion-Beam-Induced Epitaxial Recrystallization Method and Its Recent Applications
361
the two distinct regions (crystalline and amorphous) is mainly composed of dangling bonds,
and this particular kind of structural defect is responsible for the IBIEC process. Upon IBIEC
conditions (temperature + irradiation) there is a dynamic rearrangement of these dangling
bonds with annihilation in pairs which promotes the layer-by-layer planar recrystallization
toward the surface.
Fig. 8. 1 MeV He
+
RBS spectra in channeling along the [100] direction from a 90 nm a-Si
layer containing Fe before (as-implanted sample) and after (recrystallized sample)
irradiation with 600 keV Si
+
to a dose of 6x10
16
ions/cm
2
at 350 C.
Fig. 9. Bright-field TEM cross-sectional images of the as-implanted sample. (a) Overview of
the 90 nm thick a-Si layer containing Fe. (b) High-resolution of the selected region in (a)
showing in detail the c-a interface region.
Recrystallization
362
The complete recrystallization of the amorphous Si layer was also confirmed by TEM
analyses. It should be noted that for a conventional thermal annealing at the same
temperature (350 C) and time (irradiation time ~ 160 minutes) would have produced a
regrowth of only 7.7x10
-3
nm which is a negligible amount. Therefore, ion-beam irradiation
strongly enhances the kinetics of recrystallization. Figure 10 exhibits representative TEM
micrographs of the recrystallized sample. The cross-section image, such as Fig. 10a taken
along the [110]
Si
pole and slightly tilted on the zone axis, revealed the efficient a-Si regrowth
and an impurity redistribution - nanoparticles formation after the IBIEC process. Iron was
completely swept by the moving c-a interface and retained within the precipitated
narrowing layer. Despite the amount of Fe present in the recrystallized region, the quality of
the recovered crystal appears to be very good, as demonstrated by the TEM image. Three
regions regarding the nanoparticles distribution are observed: a thin region a few
nanometers thick which is closer to the surface (R
1
); a Si region about 5 nm with a small
concentration of nanoparticles (R
2
); and a layer ( 40 nm wide) with a higher concentration
of nanoparticles (R
3
).
High-resolution cross-sectional images (HRTEM) of the R
1
and R
3
regions are shown in the
insets 10b, 10c and 10d. In inset 10b, it is possible to identify small irregular shaped
nanoparticles at Si subsurface R
1
. In the deeper layers (R
2
and R
3
), two morphological
variants of the metastable -FeSi
2
phase were observed and recognized: spherical-like
nanoparticles epitaxially formed in the substrate with a fully aligned orientation regarding
the Si matrix (Fig. 10c) and plate-like nanoparticles rotated with respect to the Si matrix (Fig.
10d) as previously reported (Lin et al., 1994). The spherical-like nanoparticles form coherent
interfaces with the Si matrix, while the plate-like ones are elongated along Si( 112 )
directions.
Fig. 10. TEM images of the recrystallized sample. (a) [110]
Si
cross-sectional revealing three
nanoparticles regions in depth. HRTEM showing: (b) irregular shaped -FeSi
2
nanoparticles
at Si subsurface, (c) -FeSi
2
spherical-like and (d) -FeSi
2
plate-like nanoparticles in a deeper
region. (e) [001]
Si
SAED pattern indicating the concomitant presence of both - and -FeSi
2
phases.
Ion-Beam-Induced Epitaxial Recrystallization Method and Its Recent Applications
363
Selected-area electron diffraction (SAED) pattern obtained from a plan-view specimen is
presented in figure 10e. In addition to strong Bragg reflections of Si, extra spots due to the
nanoparticles are apparent in the diffraction pattern. The extra spots show symmetric net
patterns suggesting that there is a certain orientation relationship between the nanoparticles
and the substrate lattice. From the analysis of the extra spots symmetry and lattice spacing,
the diffraction pattern of Figure 10e can be explained as the overlap of cubic -FeSi
2
and
tetragonal -FeSi
2
phases. The reflections are consistent with
1
[112] ,
2
[11 2] (straight
lines) and [100]
(dotted line) net patterns (Behar et al., 1996; Vouroutzis et al., 2008). As the
SAED measurements have identified two crystalline phases, the near-surface precipitated
layer should contain -FeSi
2
nanoparticles.
3.2 X-ray multiple diffraction
Bragg-Surface Diffraction (BSD) (Chang, 2004) is a special diffraction case of the X-ray
multiple diffraction (XRMD) technique which has become a very useful and high resolution
probe to study in-plane effects in single crystals in general, and also, with several interesting
contributions to semiconductor epitaxial systems (dos Santos et al., 2009; Morelho &
Cardoso, 1993; Morelho et al., 1998; Orloski et al., 2005; Lang et al., 2010b). For a more
complete understanding of the experimental results that will follow in this chapter, we
briefly discuss the physical aspects of the XRMD technique.
The XRMD phenomenon is systematically generated by aligning the primary planes of a
single crystal - generally parallel to the sample surface, to diffract the incident beam and, by
rotating it around the normal to the primary (h
p
, k
p
, l
p
) planes while the diffracted beam is
monitored by a detector. Under rotation (|-axis), several other secondary (h
s
, k
s
, l
s
) planes
which are inclined with respect to the surface can enter into diffraction condition
simultaneously with the primary ones. A closer observation of the diffraction geometry
shows that other diffraction planes, the so-called coupling (h
c
, k
c
, l
c
), also interact with the
secondary diffracted beams to re-scatter them towards the detector. The obtained XRMD
pattern, called Renninger scanning (RS) (Renninger, 1937), shows a series of peaks
distributed according to the symmetry of the chosen primary vector and also to the
symmetry plane established by rotation of the several reciprocal space secondary points
when entering and leaving the Ewald sphere. Therefore, one can clearly observe in a RS,
these two types of symmetry mirrors whose position and intensity distributions are essential
for most of the applications of this technique. When a peak in the RS represents an
interaction of the incident, the primary and one secondary beam, it shows up as a three-
beam peak (or three-beam case). However, one can have two or three secondary beams
simultaneously interacting to provide four or five-beam cases (or even cases for n > 5
interacting beams) being these secondary beams either Bragg (reflected) or Laue
(transmitted) cases.
A special three-beam XRMD case, called Bragg-Surface Diffraction (BSD), appears under
adequate conditions, that is, when the secondary diffracted beam propagates along the
crystal surface under an extreme asymmetric geometry. A schematic diagram of the
multiple scattering for the BSD case occurring inside the crystal, can be seen in Figure 11,
where H
ij
are the reciprocal lattice vectors corresponding to the primary planes (H
01
),
secondary planes (H
02
) and coupling plane (H
21
). These BSD beams carry information on the
Recrystallization
364
sample surface which are useful for studying surface impurity incorporation effects (Lai et
al., 2005), or even on the interface (layer/substrate), if such an interface is present, as in the
semiconductor epitaxial structures. The technique has provided significant information
when successfully applied as a special high-resolution 3D probe for the study of epitaxial
layered heterostructures. Here, the layer and substrate lattices may be separately studied
just by selecting one adequate layer or substrate peak. The technique is also a method
capable of measuring strain fields at the interfaces in epilayer/substrate systems with depth
penetration of 2 resolution and with enough sensitivity to detect lattices distortions in the
range of 72 around the epilayer/substrate interface (Sun et al., 2006).
Fig. 11. BSD scheme using its consecutive scattering model with H
01
(primary), H
02
(secondary) and H
21
(coupling) vectors. The coupling planes re-scatter the secondary beam
towards the primary direction.
Besides the standard RS, another XRMD scanning methodology can give information on the
crystalline quality obtained from an analysis of the :| mapping scans (Morelho &
Cardoso, 1996). By using this method, the | rotation is performed for a range of e angles
each targeting an exact angular position of the multiple-beam Bragg condition. This
approach results in a three dimensional plot of the primary intensity versus e and | in a
coupled way from which, through analysis of the iso-intensity contours of such plots, one
can obtain information on the lattice coherence along the beam path and hence, on the
crystalline quality. It has been shown that when the FWHM (full width at half maximum) of
the peak in the | scan is larger in comparison to the one in the e scan, there is almost no loss
of coherence, i.e. confirming that the crystal is perfect or nearly perfect.
3.2.1 BSD reflections Results and discussion
Measurements of the (004) symmetrical high resolution rocking curves (HRRC) are
shown in figure 12 at two perpendicular orientations on the sample surface: | = 0 (Fig.
12a) and 90 (Fig. 12b). Both patterns present practically the same result with two distinct
peaks corresponding to R
2
and R
3
regions, clearly seen in each pattern, with smaller
perpendicular lattice parameters in comparison to the matrix peak (stronger). Also, (002)
HRRC were measured at ( 111 ) and (111) BSD reflections then, at two azimuth angles on
the recrystallized sample (| = 6.04 and 83.96). The results are shown in figures 12c
and 12d.
Ion-Beam-Induced Epitaxial Recrystallization Method and Its Recent Applications
365
Fig. 12. High resolution rocking curves of the recrystallized sample for (004) reflection at
| = 0 (a) and | = 90 (b) and (002) reflection at two BSD peaks: 6.04 (c) and 83.96 (d). Si
pristine is also added for comparison purposes.
As (002) is a forbidden reflection of the Si space group, no primary intensity can be observed
out of the BSD secondary peaks. The rocking obtained at | = 6.04 shows three different
contributions: the stronger peak due to the Si matrix; the broad peak to the right (higher
angles), due to the R
2
and R
3
convoluted peaks; and a distinct peak to the left (lower angles),
probably associated only with the R
3
region. It should be noticed that the rocking at | =
83.96 exhibits a meaningful profile difference, that is, the peak to the left (lower angles)
does not appear as discriminated as in the previous measurement (| = 6.04), it means, a
noticeable confirmation of the anisotropic behavior. This anisotropy, observed between the
[ 110 ] and [110] in-plane directions, could be associated with the plate-like nanoparticles
(Fig. 10d) since the shape and orientation of these ordered nanoparticles should introduce
different strains in both perpendicular directions.
Figure 13 shows the measured e:| mappings for a Si matrix (pristine) and a recrystallized
sample for comparison purposes to provide a better visualization and characterization of the
detected anisotropy. The exact BSD reflection is tailored in both e and |-directions for each
of the two above mentioned BSD secondary reflections: ( 111 ) Si pristine in Fig. 13a and
recrystallized sample in 13b and, for (111) Si pristine in 13c and recrystallized one in 13d. In
fact, these mappings give a more complete 3D view of the BSD reflection condition which
complements the 2D analysis obtained from the HRRC in Figs. 12c and 12d. Furthermore,
the mappings allow for the lattice parameters and 2D-strain determination of the distorted
regions (R
2
and R
3
). The mappings obtained for the Si matrix (pristine) along the two in-
plane perpendicular directions as depicted in 13a and 13c, exhibit only the BSD matrix peak
at = 16.578 and, as expected, no difference is observed. In turn, the recrystallization
process induces FeSi
2
nanoparticles nucleation within the implanted matrix and then, a huge
broadening as well as a striking difference is clearly observed in Figs. 13b and 13d
mappings. Besides the BSD matrix peak in Fig. 13b, two other peaks are also detected: one
Recrystallization
366
upper-side ( ~ 16.59) and one lower-side ( ~ 16.57) with respect to the matrix peak
whereas, in Fig. 13d just the matrix and the upper-side peak are clearly seen since the lower-
side peak appears as a shoulder of the matrix one. This result confirms the one obtained in
Figs. 12c and 12d.
Fig. 13. Strain anisotropy in MBSD: Si pristine (a) and recrystallized sample (b) at ( 111 ) BSD
(| = 6.04) and Si pristine (c) and recrystallized (d) at (111) BSD (| = 83.96).
Perpendicular and in-plane lattice parameters as well as strain values were obtained from
the IBIEC sample e:| mappings for | = 6.04 ( 111 ) and 83.96 (111). The perpendicular
strain obtained with the upper-side peaks of the BSD ( 111 ) and (111) e:| mappings are c
=
5.8(6)x10
-4
whereas, the in-plane strains are c
||
= 0. Then, as no anisotropy strain is
detected from this upper-side peak (e = 16.588) one can assume that most of this result can
be assigned to the spherical-like nanoparticles rather than to the plate-like ones. In turn, for
the lower-side peak, one observes that c
and c
||
values are distinct: c
= 5.4(6)x10
-4
and c
||
=
3.1(7)x10
-4
for ( 111 ) BSD peak and c
= 2.4(8)x10
-4
and c
||
= 1.3(9)x10
-4
for (111) and then, an
analogous behavior happens for a
and a
||
. Thus, one concludes there is anisotropy in the
lattice parameters and strain in both sample directions: out-of-plane and in-plane. This
anisotropy is attributed to the shape and distortion along the ( 111 ) and (111)
crystallographic planes of the plate-like nanoparticles.
4. Conclusion
The conditions for the occurrence of the recrystallization and amorphization phenomena
of a Si thin layer have been discussed in terms of the energy deposited by an ion beam as
well as the sample temperature during an irradiation process. Also, the dependence of the
recrystallization in relation to some impurity species, in particular, iron atoms dissolved
into an amorphous Si layer has been discussed. Furthermore, it was shown how the ion-
beam-induced epitaxial crystallization (IBIEC) process can be used as a method to
synthesize nanoparticles within a Si matrix. For a specific case of Fe
+
low dose implanted
Ion-Beam-Induced Epitaxial Recrystallization Method and Its Recent Applications
367
in Si(100) substrate, nanoparticles with different orientations and morphologies were
observed after the IBIEC process. These nanoparticles have caused interesting distortions
in the surrounding Si lattice. In order to structurally characterize these distortions the
Bragg Surface Diffraction was used as a 3D high-resolution tool. This non-conventional X-
ray diffraction technique was able to discriminate between the out-of-plane and in-plane
strain effects and to provide the direct observation of an important in-plane strain
anisotropy.
5. References
Behar, M.; Bernas, H.; Desimoni, J.; Lin, X. W. & Maltez, R. L. (1996). Sequential phase
formation by ion-induced epitaxy in Fe-implanted Si(001). Study of their properties
and thermal behavior. Journal of Applied Physics, Vol. 79, No. 2, (January 1996), pp.
752-762, ISSN 1089-7550
Chang, S. L. (2004). X-ray multiple-wave diffraction: theory and applications, Springer Series in
Solid-State Sciences, Vol. 143, Springer-Verlag, ISBN 3-540-21196-9, Berlin,
Germany
Dearnaley, G.; Freeman, J. H.; Nelson, R. S. & Stephen, J. (1973). Ion Implantation, Noth-
Holland Publishing Company, ISBN 0-7204-1758-9, Amsterdam, The Netherlands
Donovan, E. P.; Spaepen, F.; Turnbull, D.; Poate, J. M. & Jacobson, D. C. (1985). Calorimetric
studies of crystallization and relaxation of amorphous Si and Ge prepared by ion
implantation. Journal of Applied Physics, Vol. 57, No. 6, (March 1985), pp. 1795-1804,
ISSN 1089-7550
Donovan, E. P.; Spaepen, F.; Poate, J. M. & Jacobson, D. C. (1989). Homogeneous and
interfacial heat releases in amorphous Silicon. Applied Physics Letters, Vol. 55, No.
15, (October 1989), pp. 1516-1518, ISSN 1077-3118
dos Santos, A. O.; Lang, R.; de Menezes, A. S.; Meneses, E. A.; Amaral, L.; Reboh, S. &
Cardoso, L. P. (2009). Synchrotron x-ray multiple diffraction in the study of Fe
+
ion
implantation in Si(0 0 1). Journal of Physics D: Applied Physics, Vol. 42, No. 19,
(October 2009), pp. 195401-195407, ISSN 1361-6463
Elliman, R. G.; William, J. S.; Maher, D. M. & Brown, W. L. (1985). Kinetics, microstructure
and mechanisms of ion beam induced epitaxial crystallization of semiconductors.
Materials Research Society Symposium Proceedings, Vol. 51, (January 1985), pp. 319-
328, ISSN 0272-9172
Elliman, R. G.; Williams, J. S.; Brown, W. L.; Leiberich, A.; Maher, D. M. & Knoell, R. V.
(1987). Ion-beam-induced crystallization and amorphization of silicon. Nuclear
Instruments and Methods in Physics Research Section B Beam Interactions with Materials
and Atoms, Vol. 19-20, No. Part 2, (January 1987), pp. 435-442, ISSN 0168-583X
Holmn, G.; Linnros, J. & Svensson, B. (1984). Influence of energy transfer in nuclear
collisions on the ion beam annealing of amorphous layers in silicon. Applied Physics
Letters, Vol. 45, No. 10, (November 1984), pp. 1116-1118, ISSN 1077-3118
Jackson, K. A. (1988). A defect model for ion-induced crystallization and amorphization.
Journal of Materials Research, Vol. 3, No. 6, (November 1988), pp. 1218-1226, ISSN
2044-5326
Lai, X.; Roberts, K. J.; Bedzyk, M. J.; Lyman, P. F.; Cardoso, L. P. & Sasaki, J. M. (2005).
Structure of habit-modifying trivalent transition metal cations (Mn
3+
, Cr
3+
) in
Recrystallization
368
nearly perfect single crystals of potassium dihydrogenphosphate as examined by
X-ray standing waves, X-ray absorption spectroscopy, and molecular modeling.
Chemistry of Materials, Vol. 17, No. 16, (August 2005), pp. 4053-4061, ISSN 1520-5002
Lang, R.; Amaral, L. & Meneses, E. A. (2010a). Indirect optical absorption and origin of the
emission from -FeSi
2
nanoparticles: Bound exciton (0.809 eV) and band to acceptor
impurity (0.795 eV) transitions. Journal of Applied Physics, Vol. 107, No. 10, (May
2010), pp. 103508/1-103508/7, ISSN 1089-7550
Lang, R.; de Menezes, A. S.; dos Santos, A. O.; Reboh, S.; Meneses, E. A.; Amaral, L. &
Cardoso, L. P. (2010b). X-ray Bragg-Surface Diffraction: a tool to study in-plane
strain anisotropy due to ion-beam-induced epitaxial crystallization in Fe
+
-
implanted Si(001). Crystal Growth and Design, Vol. 10, No. 10, (August 2010), pp.
4363-4369, ISSN 1528-7505
Lin, X. W.; Washburn, J.; Liliental-Weber, Z. & Bernas, H. (1994). Coarsening and phase
transition of FeSi
2
precipitates in Si. Journal of Applied Physics, Vol. 75, No. 9, (May
1994), pp. 4686-4694, ISSN 1089-7550
Linnros, J.; Svensson, B. & Holmn, G. (1984). Ion-beam-induced epitaxial regrowth of
amorphous layers in silicon on sapphire. Physical Review B, Vol. 30, No. 7, (October
1984), pp. 3629-3638, ISSN 1550-235X
Linnros, J.; Holmn, G. & Svensson, B. (1985). Proportionality between ion-beam-induced
epitaxial regrowth in silicon and nuclear energy deposition. Physical Review B, Vol.
32, No. 5, (September 1985), pp. 2770-2777, ISSN 1550-235X
Linnros, J.; Elliman, R. G. & Brown, W. L. (1986). The competition between ion beam
induced epitaxial crystallization and amorphization in Silicon: The role of the
divacancy. Materials Research Society Symposium Proceedings, Vol. 74, (January 1986),
pp. 477-480, ISSN 0272-9172
Linnros, J.; Elliman, R. G. & Brown, W. L. (1988). Divacancy control of the balance between
ion-beam-induced epitaxial cyrstallization and amorphization in silicon. Journal of
Materials Research, Vol. 3, No. 6, (November 1988), pp. 1208-1211, ISSN 2044-5326.
Lulli, G.; Merli, P. G. & Antisari, M. V. (1987). Solid-phase epitaxy of amorphous silicon
induced by electron irradiation at room temperature. Physical Review B, Vol. 36, No.
15, (November 1987), pp. 8038-8042, ISSN 1550-235X
Miyao, M.; Polman, A.; Sinke, W.; Saris, F. W. & van Kemp, R. (1986). Electron irradiation-
activated low-temperature annealing of phosphorus-implanted silicon. Applied
Physics Letters, Vol. 48, No. 17, (April 1986), pp. 1132-1134, ISSN 1077-3118
Morelho, S. L. & Cardoso, L. P. (1993). Analysis of interfacial misfit dislocation by X-ray
multiple diffraction. Solid State Communications, Vol. 88, No. 6, (November 1993),
pp. 465-469, ISSN 0038-1098
Morelho, S. L. & Cardoso, L. P. (1996). X-ray multiple diffraction phenomenon in the
evaluation of semiconductor crystalline perfection. Journal of Applied
Crystallography, Vol. 29, No. 4, (August 1996), pp. 446-456, ISSN 1600-5767
Morelho, S. L.; Avanci, L. H.; Hayashi, M. A.; Cardoso, L. P. & Collins, S. P. (1998).
Observation of coherent hybrid reflection with synchrotron radiation. Applied
Physics Letters, Vol. 73, No. 15, (October 1998), pp. 2194-2196, ISSN 1077-3118
Ion-Beam-Induced Epitaxial Recrystallization Method and Its Recent Applications
369
Narayan, J.; Fathy, D.; Oen, O. S. & Holland, O. W. (1984). High-resolution imaging of ion-
implantation damage and mechanism of amortization in semiconductors. Materials
Letters, Vol. 2, No. 3, (February 1984), pp. 211-218, ISSN 0167-577X
Olson, G. L. & Roth, J. A. (1988). Kinetics of solid phase crystallization in amorphous silicon.
Materials Science Reports, Vol. 3, No. 1, (May 1988), pp. 1-77, ISSN: 0920-2307
Orloski, R. V.; Pudenzi, M. A. A.; Hayashi, M. A.; Swart, J. W. & Cardoso, L. P. (2005). X-ray
multiple diffraction on the shallow junction of B in Si(0 0 1). Journal of Molecular
Catalysis A: Chemical, Vol. 228, No. 1-2, (March 2005), pp. 177-182, ISSN 1381-1169
Priolo, F.; La Ferla, A. & Rimini, E. (1988). Ion-beam-assisted growth of doped Si layers.
Journal of Materials Research, Vol. 3, No. 6, (November 1988), pp. 1212-1217, ISSN
2044-5326
Priolo, F.; Spinella, C.; La Ferla, A.; Rimini, E. & Ferla, G. (1989). Ion-assisted
recrystallization of amorphous silicon. Applied Surface Science, Vol. 43, No. 1-4,
(December 1989), pp. 178-186, ISSN 0169-4332
Priolo, F. & Rimini, E. (1990). Ion-beam-induced epitaxial crystallization and amorphization
in silicon. Materials Science Reports, Vol. 5, No. 6, (June 1990), pp. 319-379, ISSN:
0920-2307
Priolo, F.; Spinella, C. & Rimini, E. (1990). Phenomenological description of ion-beam-
induced epitaxial crystallization of amorphous silicon. Physical Review B, Vol. 41,
No. 8, (March 1990), pp. 5235-5242, ISSN 1550-235X
Renninger, M. (1937). Umweganregung, eine bisher unbeachtete
Wechselwirkungserscheinung bei Raumgitterinterferenzen. Zeitschrift fr Physik A
Hadrons and Nuclei, Vol. 106, No. 3-4, (July 1937), pp. 141-176
Roorda, S.; Doorn. S.; Sinke, W. C.; Scholte, P. M. L. O. & Van Loenen, E. (1989). Calorimetric
Evidence for Structural Relaxation in Amorphous Silicon. Physical Review Letters,
Vol. 62, No. 16, (April 1989), pp. 1880-1883, ISSN 1079-7114
Sun, W. C.; Chang, H. C.; Wu, B. K.; Chen, Y. R.; Chu, C. H.; Chang, S. L.; Hong, M.; Tang,
M. T. & Stetsko, Y. P. (2006). Measuring interface strains at the atomic resolution in
depth using x-ray Bragg-surface diffraction. Applied Physics Letters, Vol. 89, No. 9,
(August 2006), pp. 091915/1-091915/3, ISSN 1077-3118
Vouroutzis, N.; Zorba, T. T.; Dimitriadis, C. A.; Paraskevopoulos, K. M.; Dzsa, L. & Molnr,
G. (2008). Growth of -FeSi
2
particles on silicon by reactive deposition epitaxy.
Journal of Alloys and Compounds, Vol. 448, No. 1-2, (January 2008), pp. 202-205, ISSN
0925-8388
Washburn, J.; Murty, C. S.; Sadana, D.; Byrne, P.; Gronsky, R.; Cheung, N. & Kilaas, R.
(1983). The crystalline to amorphous transformation in silicon. Nuclear Instruments
and Methods in Physics Research, Vol. 209-210, No. Part 1, (May 1983), pp. 345-350,
ISSN 0167-5087
Williams, J. S. (1983). Solid phase recrystallisation process in Silicon, In: Surface modification
and alloying by laser, ion and electron beams, Poate, J. M.; Foti, G.; Jacobson, D. C.,
pp. 133, Plenum Press, ISBN 0306413736, New York, United States of America
Williams, J. S.; Elliman, R. G.; Brown, W. L. & Seidel, T. E. (1985). Dominant Influence of
beam-induced interface rearrangement on solid-phase epitaxial crystallization of
amorphous silicon. Physical Review Letters, Vol. 55, No. 14, (September 1985), pp.
1482-1485, ISSN 1079-7114
Recrystallization
370
Williams, J. S. & Elliman, R. G. (1986). Role of electronic processes in epitaxial
recrystallization of amorphous semicontluctors. Physical Review Letters, Vol. 51, No.
12, (September 1983), pp. 1069-1072, ISSN 1079-7114
Ziegler, J. F. (2011). Stopping and range of ions in matter, SRIM-2011. Available from
http://www.srim.org/
0
Steady-State Grain Size in Dynamic
Recrystallization of Minerals
Ichiko Shimizu
Department of Earth and Planetary Science, University of Tokyo, Tokyo
Japan
1. Introduction
Dynamic recrystallization (DRX) is a strain restoration and grain renement mechanism that
occurs in high-temperature dislocation creep of metals and minerals (Humphreys & Hatherly,
2004). Microstructures indicative of DRX are commonly observed in rock-forming minerals
that have been subjected to natural deformation in the Earths crust and mantle (Fig. 1).
Laboratory studies have revealed that the average size d of recrystallized grains approaches a
steady-state value, which is determined by the applied stress and is independent of the initial
grain size. Twiss (1977) proposed a stressgrain size relation of the following form:
d
b
= K
_
_
p
(1)
where is the ow stress, is the shear modulus, b is the length of the Burgers vector, and
K is a non-dimensional constant. The grain size exponent p ranges between 1 and 1.5 for
most materials. Empirically determined d relations of minerals have been used to estimate
the stress states in the Earths interior. However, detailed studies of a Mg alloy (De Bresser
et al., 1998) and NaCl (Ter Heege et al., 2005) revealed that K has a weak dependence on
temperature. Derby & Ashby (1987) modeled the DRX processes of metals and predicted the
temperature dependence of the recrystallized grain size, but they failed to account for the
observed range of exponent p (Derby, 1992; Shimizu, 2011).
In this chapter, we focus on deformation and recrystallization processes in minerals and
examine the effects of stress and temperature on the steady-state grain size.
2. Recrystallization mechanisms in minerals
DRX was rst observed in hot deformation of cubic metals such as Cu, Ni, and austenitic
iron. A simplied description of DRX in these metals is as follows. Strain-free new grains are
usually formed by bulging of pre-existing grain boundaries and they grow at the expense of
old grains to reduce the dislocation energy of the material (Sakai, 1989; Sakai & Jones, 1984).
As the dislocation density of the new grains increases, they cease to grow and new nucleation
events occur at their margins. These processes repeat cyclically during dislocation creep.
15
2 Recrystallization
Fig. 1. Optical micrograph of a thin section of a quartz schist (Sanbagawa metamorphic belt,
Japan) under polarized transmitted light with a sensitive color plate. Blue and red represent
the orientation of the crystallographic c-axis of quartz grains. New small grains form at the
margins and interiors of larger grains.
In contrast to the classical view of DRX described above, syndeformational recrystallization
of minerals such as quartz, calcite, and olivine proceeds with progressive misorientation of
subgrain boundaries (Poirier, 1985). Subgrain rotation (SGR) recrystallization also occurs in
some metals such as Mg and Al alloys and is termed continuous DRX, whereas DRX in the
original sense is currently referred to as discontinuous DRX (Humphreys & Hatherly, 2004).
At low temperatures (T) and high strain rates ( ), SGR is localized at grain margins (Hirth
& Tullis, 1992; Schmid et al., 1980); however, intracrystalline SGR becomes more important
and grain boundary migration (GBM) occurs at high T and low (Hirth & Tullis, 1992; Rutter,
1995) (Fig. 2). Consequently, the recrystallized grain size is much larger than the subgrain size
(Guillop & Poirier, 1979; Karato et al., 1980).
For both discontinuous and continuous DRX, grain size reduction occurs at nucleation events,
whereas strain-induced GBM leads to overall coarsening. The steady-state grain size is
determined by the dynamic balance between nucleation and grain growth (Derby & Ashby,
1987).
3. Grain size distribution
In the d relation (Eq. 1), the steady-state microstructure is represented by a single value of
the average grain size d, but dynamically recrystallized materials generally have wide grain
size distributions. As a simplied model of DRX, Shimizu (1998a; 1999; 2003) considered
following nucleation and growth processes and analyzed the evolution of the grain size
distribution:
1. Nucleation occurs at a constant rate I per unit volume.
2. Nucleation sites are randomly distributed.
3. Each grain grows with a radial growth rate
R.
372 Recrystallization
Steady-State Grain Size in Dynamic Recrystallization of Minerals 3
Fig. 2. Nucleation and growth in continuous DRX. Solid lines represent grain boundaries and
thin dotted lines represent subgrain boundaries. Nucleated grains (yellow) are formed by
SGR and grow in the deformed matrix.
4. Newly crystallized grains replace older grains.
In the steady state, the grain size has a nearly a log-normal distribution and many newly
crystallized grains coexist with a few old grains in a certain population balance. The average
grain size satises
d = a
_
R
I
_
1
4
(2)
where a is a scaling factor; a = 1.14 for a 3D distribution and a = 1.12 for a distribution
measured in a 2Dsection. Shimizu (1998b; 2008; 2011) considered strain-induced grain growth
for
R (Sec. 4) and SGR nucleation for I (Sec. 5) and derived the d relation for continuous
DRX (Sec. 6). In Sec. 7, we revise the theoretical model to incorporate the inuence of the
surface-energy drag.
4. Strain-induced grain growth
4.1 Dislocation energy
During high-T dislocation creep of minerals, dynamic recovery cooperates with continuous
DRX and assists subgrain formation. Unrecovered microstructures such as tangled
dislocations are rarely observed in recrystallized grains (Hirth & Tullis, 1992). Hence,
the strain energy (E
strain
) is given by a sum of the energies of isolated dislocations and
sub-boundaries (E
disl
and E
sub
, respectively):
E
strain
= E
disl
+ E
sub
(3)
373 Steady-State Grain Size in Dynamic Recrystallization of Minerals
4 Recrystallization
The free dislocation energy per unit volume is
E
disl
= (4)
where is the dislocation density and is the dislocation line tension. When the internal
stress around dislocations is equilibrated with the applied stress , the following equation
holds (Nabarro, 1987):
= b
1
2
(5)
where is a constant that depends on the conguration of the dislocation arrays. Hence,
=
_
b
_
2
(6)
The dislocation line tension is given by (Hirth & Lothe, 1982)
=
b
2
4
ln
_
r
b
_
(7)
where r is the characteristic radius of the elastic eld around the dislocation core and the
constant is typically in the range 34. The parameter depends on the dislocation
conguration:
1
2
(9)
Substituting Eqs. (6) and (9) into Eq. (7) yields
=
b
2
4(1 )
ln
_
2
_
(10)
Substituting Eq. (10) into Eq. (4) and using Eq. (6) again, we have
E
disl
=
2
4
2
(1 )
ln
_
2
_
(11)
4.2 Sub-boundary energy
Consider nearly spherical subgrains with a diameter d
1
2
=
3
d
(13)
The factor 1/2 is included because the area of each subgrain wall is counted twice. The energy
of sub-boundaries in a unit volume of the material can thus be written as
E
sub
=
3
d
(14)
where is the sub-boundary energy per unit area.
The theory of dislocations gives
=
b
2
4(1 )h
() (15)
() coth ln(2 sinh ) (16)
b
h
(17)
where h is the mean dislocation spacing (Hirth & Lothe, 1982).
For a tilt boundary (Fig. 3), h and the misorientation angle are related by (Poirier, 1985)
b
h
= 2 tan
_
2
_
(18)
The last approximation is justied for low-angle boundaries. Then, Eq. (17) becomes
=
1 (19)
Hence, the following approximations can be applied to Eq. (16):
coth
1
, sinh (20)
Then, Eq. (15) becomes
=
2
b (21)
where
1
2(1 )
_
1 ln
_
2
__
(22)
The subgrain size is empirically expressed as (Takeuch & Argon, 1976; Twiss, 1977)
d
b
= K
_
1
(23)
375 Steady-State Grain Size in Dynamic Recrystallization of Minerals
6 Recrystallization
where K
(24)
4.3 Subgrain size
We consider a recovery process in which free dislocations with a dislocation density
rearrange into sub-boundaries. Conservation of the total dislocation length during subgrain
formation requires
=
A
h
(25)
The right-hand side represents the length of dislocations in sub-boundaries. Using Eqs. (13)
and (18), the above expression is modied to become
=
3
d
b
(26)
Subgrains are formed if the total sub-boundary energy is smaller than the free dislocation
energy (Twiss, 1977):
E
disl
E
sub
(27)
The equality represents the critical state for the initiation of subgrain formation. From Eqs. (4)
and (14), this condition can be written as
3
d
(28)
h
h
b
Fig. 3. Schematic illustration of a tilt boundary with a misorientation angle , dislocation
spacing h, and Burgers vector b.
376 Recrystallization
Steady-State Grain Size in Dynamic Recrystallization of Minerals 7
Substituting Eqs. (6), (7), and (21)(22) into the above expression for , , and , respectively,
Eq. (28) becomes
_
b
_
2
b ln
_
2
_
3
d
_
1 ln
_
2
__
(29)
Equating Eqs. (26) and (6), we have
_
b
_
2
=
3
d
b
(30)
Then, Eq. (29) reduces to
ln
_
2
_
_
1 ln
_
2
__
(31)
The stability limit of is derived as
e
_
(32)
where e is the Napierian base. The equality gives the initial misorientation angle
i
:
i
=
e
_
(33)
Applying
i
to of Eq. (30), the initial subgrain size d
i
is obtained as
d
i
b
=
3e
_
1
(34)
Once the subgrain boundary is established, it functions as a dislocation sink because
progressive subgrain misorientation is an energetically favorable process. We thus assume
that the subgrain size is maintained during the subsequent misorientation. Substituting
d
= d
i
into Eq. (34), we obtain Eq. (23), where
K
=
3e
(35)
Using Eqs. (22) and (35), the full expression of Eq. (24) is obtained as
E
sub
=
2e
=
4e(1 )
_
1 ln
_
2
__
(36)
4.4 Growth kinetics
The kinetic law of grain growth is generally written as
R = MF (37)
where M is the mobility of the grain boundary and F is the driving force. M depends on T as
M =
bwD
gb
kT
(38)
377 Steady-State Grain Size in Dynamic Recrystallization of Minerals
8 Recrystallization
0 200 400 600 800
0
1000
2000
3000
4000
[MPa]
E
n
e
r
g
y
[
k
J
/
m
3
]
Quartz
misorientation angle = 2
E
disl
E
sub
E
strain
Fig. 4. Strain energy of quartz calculated using Eqs. (11) and (36). The physical parameters of
quartz are given as (Shimizu, 2008) = 3 (Kohlstedt & Weathers, 1980), = 4.2 10
4
MPa,
and = 0.15 (Twiss, 1977). Because no data are available for of quartz, we apply = 3 of
ionic crystals (Hirth & Lothe, 1982).
D
gb
= D
gb
exp
_
Q
gb
RT
_
(39)
where w is the boundary width, k is the Boltzmann constant, D
gb
is the diffusion coefcient at
the grain boundary, D
gb
is a constant, R is the gas constant, and Q
gb
is the activation energy
for grain boundary diffusion.
In a single-phase material, grain growth occurs to reduce the bulk strain energy and the energy
of grain surfaces. Hence, Eq. (37) is written as
R = M(F
strain
+ F
sur f
) (40)
where F
strain
and F
sur f
represent the driving forces due to strain energy and surface energy
(grain boundary energy), respectively. The strain energy in dynamically recrystallized
materials is not homogeneous. The strain energy of deformed grains is given by the sum
of E
disl
in Eq. (11) and E
sub
in Eq. (36), whereas newly recrystallized grains are almost strain
free. This difference in strain energy drives grain growth. Hence,
F
strain
= E
strain
(41)
With increasing strain, free dislocations multiply and excess dislocations rearrange into
sub-boundaries. Then, increases and the sub-boundary energy exceeds the free dislocation
energy. Fig. 4 shows the calculations for quartz. When the average misorientation angle
reaches several degrees, the following approximation can be used instead of Eq. (3):
E
strain
E
sub
(42)
378 Recrystallization
Steady-State Grain Size in Dynamic Recrystallization of Minerals 9
5. Nucleation rate
In SGR nucleation, the nuclei are approximately the same size as the original subgrains. Thus,
the number of potential nucleation sites per unit volume of crystals is given by Eq. (12) for
intracrystalline nucleation and
N =
6
dd
2
(43)
for nucleation at grain margins (Fig. 2b). The nucleation rate is scaled as
I =
N
c
(44)
where
c
is the interval of nucleation events.
The subgrain becomes a nucleus when the misorientation angle exceeds a critical value
c
. The ux of dislocations that move toward the sub-boundary is given by u, where u is
the climb velocity. The time required for dislocations to accumulate at the sub-boundary is
equal to the nucleation cycle
c
. From Eq. (18), a critical nucleus has a dislocation spacing of
h
c
= b/
c
; hence, the number of dislocations per unit area of the boundary is 1/h
c
=
c
/b.
Dividing this value by the ux u, the nucleation cycle is evaluated as
c
c
bu
(45)
The climb velocity of dislocations is given by (Hirth & Lothe, 1982)
u =
D
v
lkT
(46)
where is the atomic volume, D
v
is the self-diffusion coefcient, and l is a length scale given
by
l
b
2
ln
_
r
b
_
(47)
Using Eqs. (6) and (9), Eq. (47) can be rewritten as
l =
b
2
ln
_
2
_
(48)
The temperature dependence of D
v
is expressed as
D
v
= D
v
exp
_
Q
v
RT
_
(49)
where D
v
is a constant and Q
v
is the activation energy for volume diffusion.
Combining Eqs. (44)(46), approximating as b
3
, and using Eqs. (12) and (23), we have
I =
6
bK
3
c
D
v
lkT
_
_
5
(50)
379 Steady-State Grain Size in Dynamic Recrystallization of Minerals
10 Recrystallization
for intracrystalline nucleation. Using Eq. (43) instead of Eq. (12), the equation for marginal
nucleation is obtained:
I =
1
d
6
K
2
c
D
v
lkT
_
_
4
(51)
6. Scaling relation
Here, we neglect the surface energy term in Eq. (40) and assume
F = MF
strain
(52)
Combining Eq. (2) with Eqs. (52), (38), (39), (41), and (42), and using Eq. (36) and either Eq.
(50) or Eq. (51), the steady-state grain size in continuous DRX is derived as
d
b
= B
_
_
p
_
wD
gb
bD
v
_
1
m
(53)
where
p =
5
4
= 1.25, m = 4 (54)
for intracrystalline nucleation and
p =
4
3
= 1.33, m = 3 (55)
for marginal nucleation (Shimizu, 1998b; 2008). B is a non-dimensional constant given by
B =
_
a
4
K
m2
2
4
c
l
b
_
1/m
(56)
Using Eq. (48), Eq. (56) can be rewritten as
B =
_
a
4
K
m2
c
8
ln
_
2
_
_
1/m
(57)
Although is included in the right-hand side, the stress dependence of B is negligibly small.
Using Eqs. (39) and (49), Eq. (53) can be re-expressed as
d
b
= K
_
p
exp
_
Q
mRT
_
(58)
where
K
= B
_
wD
gb
bD
v
_
1/m
(59)
and
Q = Q
gb
Q
c
(60)
380 Recrystallization
Steady-State Grain Size in Dynamic Recrystallization of Minerals 11
d
F
r
e
q
.
d'
F
strain
(a)
d
F
r
e
q
.
d'
F
surf
(b)
Fig. 5. Schematic representation of grain size evolution due to (a) strain-energy-driven grain
growth and (b) surface-energy drag.
As Q
gb
is generally smaller than Q
c
, the recrystallized grain size is predicted to have a weak
positive dependence on T. The constant K in Eq. (1) can now be written as a function of T:
K = K
exp
_
Q
mRT
_
(61)
7. Inuence of surface energy
We now consider the inuence of surface energy (grain boundary energy). In the case
of surface-energy-driven grain coarsening in single-phase materials under static conditions
(known as normal grain growth), large grains are energetically favorable and grow at the
expense of small grains; the evolution of individual grain size has the opposite sense to that
considered for DRX in Sec. 3 (Fig. 5). Therefore, when new grains grow by the strain-energy
difference, the surface energy acts as a drag force.
In the theory of normal grain growth (Hillert, 1965), grain size evolution is described by
R
k
= Mc
_
1
R
1
R
k
_
(62)
where R
k
and
R
k
are respectively the radius and the growth rate of the k-th grain and c 1 is a
statistical factor. If R
k
is smaller (larger) than the mean radius R, the above expression becomes
381 Steady-State Grain Size in Dynamic Recrystallization of Minerals
12 Recrystallization
negative and the k-th grain shrinks (grows). By comparison with Eq. (37), the driving force for
the growth of the k-th grain can be written as
2c
_
1
d
1
d
k
_
(63)
where d
k
is the diameter of the k-th grain. In the nucleation and growth processes in DRX, the
inuence of the surface-energy drag is largest for small nuclei. Thus, we introduce a modied
factor c
_
1
d
1
d
_
2c
1
d
(64)
With this equation and Eq. (42), Eq. (40) can be approximated as
R M
_
E
sub
2c
_
(65)
Using this equation, Eq. (56) can be modied as follows (the parameters p, m, and Q remain
the same).
B =
_
a
4
K
m2
c
8
_
3
2
2c
b
_
ln
_
2
_
_
1/m
(66)
8. Comparison of theory with experiments
8.1 Stress dependence of recrystallized grain size
In Fig. 6, p values of rock-forming minerals determined by triaxial or uniaxial or compression
tests are plotted against the n-th power of dislocation creep ow laws (
n
), which reect
the rate-controlling processes of dislocation creep; for climb-controlled creep, n is generally
35. The gure also shows the experimental result for a hexagonal Mg alloy (Magnox Al80),
which was studied as a quartz analogue (De Bresser et al., 1998). The observed p values are
almost independent of the power-law exponents and are well explained by the present model
for continuous DRX.
8.2 Application to quartz
The theoretical model for the recrystallized grain size was applied to quartz using the
equations presented in Sec. 6 (Shimizu, 2008; 2011). However, the previous model accounted
only for strain energy; it neglected the effects of surface energy. Moreover, it turned out
that the previous calculation involved a numerical error; when this error is corrected, the
theoretical d lines (Fig. 8 of Shimizu (2008)) shift to higher . Here, we recalculate the d
relation of quartz using the revised equations in Sec. 7.
Because experimentally deformed quartzite samples exhibit intracrystalline SGR at moderate
stresses (Hirth & Tullis, 1992; Stipp & Tullis, 2003), we apply the intracrystalline nucleation
model (Eq. 54). In addition to the material constants given in the caption of Fig. 4, we use
b = 5 10
4
m(Twiss, 1977), = 2
,
c
= 12
, and D
v
and D
gb
of oxygen in -quartz (Farver
& Yund, 1991b; Giletti & Yund, 1984). For grain boundary energy, we use = 0.27 Jm
2
382 Recrystallization
Steady-State Grain Size in Dynamic Recrystallization of Minerals 13
3 4 5 6 7 8
0.2
0.6
1.0
1.4
1.8
n
p
IN
MN
Quartz
Olivine
Calcite
Halite
Mg-alloy
1
2
3
5
6
4 3
7
Fig. 6. Stress exponent p of recrystallized grain size plotted against the power-law exponent
n of dislocation creep (IN: intracrystalline nucleation model; MN: marginal nucleation
model). 1: Stipp & Tullis (2003) for p, Gleason & Tullis (1995) for n, 2: Karato et al. (1980) for
p, Karato et al. (1986) for n, 3: van der Wal et al. (1993) for p, Chopra & Paterson (1984) for n,
4: Rutter (1995) for p, Schmid et al. (1980) for n, 5: Guillop & Poirier (1979) for p and n, 6: Ter
Heege et al. (2005) for p, Carter et al. (1993) for n, 7: De Bresser et al. (1998) for p and n.
Microstructures of SGR and GBM are reported from all experiments except Ref. 3.
(Hiraga et al., 2007) and assume c
1.25
exp
_
7.25 kJ/mol
RT
_
; quartz (67)
In this expression, the weak stress dependence of B in Eq. (66) is neglected and B = 1.01 at
=50 MPa is chosen as a representative value. The calculation results (Fig. 7a) agree well with
the empirical data for -quartz (Stipp & Tullis, 2003). For comparison, the d relation based
on the marginal nucleation model is also shown.
In Fig. 7(b), the theoretical model is extended to the -quartz stability eld in which D
v
of
oxygen in -quartz (Farver & Yund, 1991a) is used and - and -quartz are assumed to have
the same Q
v
/Q
gb
ratio. The recrystallized grain size of -quartz is predicted to be
d = 9.98 10
2
1.25
exp
_
12.4 kJ/mol
RT
_
; quartz (68)
With decreasing temperature, the steady-state grain size shifts to higher stresses. If the
empirical d relation is directly applied to natural rocks that have deformed under low-T
( 400
m
]
10 100 1000
1
10
100
Intracrystalline
-quartz
-quartz
nucleation model
1000
C
800
C
600
C
500
C
400
C
300
C
10 100 1000
1
10
100
-quartz
[MPa]
d
[
m
]
Intracryst. nucl. model
Marginal nucl. model
Stipp & Tullis (2003)
10001100
C
1050
C
(a)
(b)
Fig. 7. Recrystallized grain size of quartz. (a) Theoretically calibrated d relations for
-quartz at 1050
C and the experimental results of Stipp & Tullis (2003). Solid line:
intracrystalline nucleation model. Dotted line: marginal nucleation model. Solid circles:
recrystallized grain size at 10001100
C) and -quartz
(red lines, 500300
(1)
with m
T
, the total mass.
Obviously, this mass is always lower than the total mass of enantiomeric excess since
e.e.(I
1
) 0.
It is then possible to proceed to a recrystallization of a pure enantiomer in the case of a
racemic compound-forming system. However, this is only applicable to a starting system
presenting a sufficient enantiomeric excess. Thus, the process requires a first enrichment
step that can be performed by chiral chromatography or via diastereoisomer formation.
Moreover, it is not possible to recover the totality of the enantiomeric excess.
V
1
R
S
RS
H
1
H
1
I
1
I
1
M
K
1
3
V
1
R
S
RS
H
1
H
1
I
1
I
1
M
K
1
3
Triphasic domains
Region 1 : Invariant liquid I1 doubly saturated + <S> + <RS>
Region 1: Invariant liquid I1 doubly saturated + <R> + <RS>
Biphasic domains:
Region 2: Saturated solution + <RS>
Region 3: Saturated solution + <S>
Region 3: Saturated solution + <R>
Monophasic domain:
Region 4: Undersaturated solution
Fig. 3. Ternary phase diagram observed for a racemic compound-forming system
T = T
1
= Cste
Recrystallization of Enantiomers from Conglomerates
409
3.2 Enantiopurification from a conglomerate-forming system
On figure 4 is depicted the ternary phase diagram observed for a conglomerate forming
system (S and R enantiomers, solvent V
1
) at temperature T
1
. This diagram is simpler that the
one corresponding to the racemic compound-forming system since only two solid phases (S
and R) can coexist. Four regions are delimited by the solubility curves (see legend of figure 4).
Here, as for the racemic compound, starting from a mixture M of enantiomers, the process
consists in adding the exact quantity of solvent (at T
1
) so that the overall synthetic mixture is
represented by point K. Point K is situated on the tie-line SI separating the 3 phase domain I-
S-R (which contains: pure <S>, pure <R> and the doubly saturated solution I) and the
biphasic domain S-H-I (where the phases in equilibrium are pure <S> and the saturated
solution). Then the selective crystallization can be pursued.
Contrary to the previous case, the whole enantiomeric excess of mixture M can be retrieved
(on the condition that the compound crystallizes as a conglomerate without partial solid
solution). It constitutes the best situation to obtain an efficient recrystallization in
thermodynamic equilibrium.
In practice (see figure 5), it is better to heat the system above T
1
so that the suspension at
T
2
> T
1
contains only pure <S> and a saturated solution. The smooth cooling program will
mainly consist in crystal growth leading to a better filterability and a more efficient washing
of the filtration cake.
It is even sometime possible to retrieve more than the initial enantiomeric excess by cooling
the system at T
0
. Point K is then situated in the 3-phase domain and as long as the
crystallization is stereoselective the liquid composition can evolve from I
1
at (T
1
) to Z
0
at T
0
(figure 6). Note that Z
0
is located on the metastable solubility curve of S at T
0
. This
corresponds to an entrainment (i.e. a single operation of the so-called AS3PC process).
V
1
R S
H
1 J
1
I
1
M
K
2
V
1
R S
H
1 J
1
I
1
M
K
2
Triphasic domain: 1 - Invariant liquid I doubly saturated + <S> + <R>
Biphasic domains: 2 - Saturated solution + <S>; 2 - Saturated solution + <R>
Monophasic domain: 3 - Undersaturated solution
Dashed lines stand for metastable solubilities
Fig. 4. Ternary phase diagram observed for a conglomerate forming system
Recrystallization
410
V
1
R S
H
1
J
1
I
1
I
2
M
K
H
2
J
2
T
1
T
2
V
1
R S
H
1
J
1
I
1
I
2
M
K
H
2
J
2
T
1
T
2
Fig. 5. Ternary phase diagrams observed for a conglomerate-forming system at T
1
(Blue) and
T
2
(Red)
V
1
R S
H
1 J
1
I
1
I
0
M
K
H
0
J
0
T
1
T
0
Z
0
H
0
J
0
K
Z
0
V
1
R S
H
1 J
1
I
1
I
0
M
K
H
0
J
0
T
1
T
0
Z
0
H
0
J
0
K
Z
0
Fig. 6. Principle of crystallization by entrainment
When the initial mixture M corresponds already to a high enantiomeric excess, solvent V
1
at
temperature T
1
wont be appropriate because the slurry will be simply much too viscous.
Thus, it is necessary to find another solvent V
2
in which the pure enantiomer <S> is poorly
soluble (figure 7).
3.3 Enantiopurification from conglomerate-forming derivatives
Figure 8 shows that a conglomerate-forming solvate (here a monohydrate) can also give the
full discrimination in the solid state even if a stable racemic compound exists for the
anhydrous chiral components. The enantiopurification by means of crystallization can be
operated in a similar way as that presented above (point K for a mixture M at T
1
). When the
Recrystallization of Enantiomers from Conglomerates
411
initial point M is close to the pure enantiomer (e.g. S), the medium can be changed by
mixing an anti-solvent. The second solvent must be miscible with water, should not induce a
miscibility gap and should not inhibit the formation of the conglomerate.
V
1
R
S
H
1
J
1
I
1
M
K
1
V
2
R S
H
2 J
2
I
2
M
K
2
H
2
J
2
Z
0
K
2
V
1
R
S
H
1
J
1
I
1
M
K
1
V
2
R S
H
2 J
2
I
2
M
K
2
H
2
J
2
Z
0
K
2
Fig. 7. Enantiomeric purification with a high enantiomeric excess Choice of another solvent
H
2
O
R
S
RS
S-H
2
O
M
K
R-H
2
O
I
1
3
4 4
5
H
2
O
R
S
RS
S-H
2
O
M
K
R-H
2
O
I
1
3
4 4
5
1 : <S-H2O> + <S> + <RS> ; 1: <R-H2O> + <S> + <RS>
2: <S-H2O> + <R-H2O> + <RS>
3: Invariant liquid I doubly saturated + <S-H2O> + <R-H2O>
4: Saturated solution + <S-H2O>
4: Saturated solution + <R-H2O>
5: Undersaturated solution
Fig. 8. Enantiomeric purification from a derivative (case of a monohydrate)
Recrystallization
412
3.4 Benefits of conglomerate
It is now obvious that proceeding to the enantiopurification with a starting product
crystallizing as a conglomerate is even more beneficial since:
- an optimized recrystallization can virtually be carried out without any loss of
enantiomeric excess which is not possible with a racemic compound.
- the choice of appropriate solvent and crystallization conditions usually allows to reach
an e.e greater than 99.5 %.
- an optimized recrystallization can be carried out on derivatives such as hydrates
(crystallizing as conglomerates) even if the molecule of interest crystallizes as a stable
racemic compound.
Note also, that when coupled to racemization, almost pure enantiomer can be obtained by
attrition from the racemic mixture (Levillain et al, 2009).
4. Detection of conglomerate
To perform preferential crystallization, the racemic mixture should crystallize as a stable
conglomerate. In most cases, the chiral resolution can not be envisaged on the original
compound because it does not fulfill this condition. An interesting alternative is to proceed
to the resolution on derivatives such as salts, co-crystals, hydrates or solvates that crystallize
as conglomerates (at the end of the resolution process, the pure enantiomer can be easily
recovered by salting out, dehydration or desolvation). To multiply the chances of spotting a
conglomerate, experiments have to be performed on a large number of non chiral acids (for
a chiral base) with various stoichiometries and solvents, with different co-crystal formers
and under various crystallization conditions.
Consequently, starting from the racemic mixture it is often necessary to synthesize series of
derivatives that have to be analyzed in order to isolate at least one conglomerate forming
system. Due to the low occurrence of conglomerates, it usually implies to investigate a large
number of derivatives.
Conglomerate detection which is an essential step of the resolution method can be pursued
by various methods which are described and discussed in the following.
4.1 Classical screening of conglomerates
4.1.1 Comparisons between IR, Raman, solid state NMR or XRPD patterns
Several spectroscopic techniques can be used to differentiate the racemic mixture and the
conglomerate via comparison between the racemic pattern and that of the enantiomer. In
case of conglomerate without partial solid solution, the patterns of the pure enantiomer
obtained from IR, Raman, solid state NMR or XRPD should be superimposable to the ones
obtained for the racemic mixture.
Even if Raman spectroscopy and IR spectroscopy can be of a certain help in spotting a
conglomerate, they cannot be considered as totally reliable. Indeed, vibrational bands
mainly arise from molecular vibrations. Only low frequency vibrations are directly related
to the vibrations of the crystal lattice. If the neighborhood of a given molecule can generate
Recrystallization of Enantiomers from Conglomerates
413
variations in the high frequency domain, two close structures can generate similar spectra.
Considering these elements, the perfect match of the XRPD patterns is surely the most
dependable way to conclude on the conglomerate nature of a sample. XRPD is of particular
interest as it also permits to check the crystallinity of the sample and to spot partial or total
solid solutions (Wermester et al, 2007; Renou et al, 2007). Moreover, if a single crystal of
sufficient size is available, a structure can be resolved by single crystal X-Ray diffraction.
The knowledge of the crystal structure definitively confirms the conglomerate nature (or
not) of the sample.
4.1.2 Alternatives methods
Other methods consist of isolating a unique particle (single crystal) from the racemic
mixture to proceed to the analysis. This particle can be dissolved in a nematic phase (Jacques
et al, 1994) or analyzed by chromatography (Chiral CG, HPLC) (Pirzada et al, 2010) and/or
polarimetry. However, there are serious limitations to these techniques since the collection
of an isolated particle can be difficult and also because the analysis of a single particle
necessitates high detection levels. To be exhaustive, let us also mentioned that the detection
of a conglomerate can also be performed via thermal analysis (the construction of the binary
phase diagram by simple measurements of the melting temperatures of the enantiomers and
that of their corresponding racemic mixture can be used to identify the nature of the sample
(conglomerate or racemic compound)(Neau et al, 1993) or Solid state circular dichrosm (SS-
CD) and CD microscopy (Kuroda et al, 2000; Claborn et al, 2003). The construction of a
ternary isotherm is also a reliable method especially when a conglomerate of solvated
phases is suspected.
4.2 High throughput technique for conglomerate prescreening
Unfortunately and despite technical improvements, all the previous methods are time
consuming and often require the pure enantiomer to be available at an early stage of the
process. Consequently, there is a need for a faster (and cheaper) method compatible with a
combinatorial approach of crystallization derivatives (salts, hydrates, solvates, cocrystals,
etc.)
Considering the data summarized in table 1, it appears that racemic compounds crystallize
not often in non centrosymmetric structures (less than 5% for NA + NC). Consequently, a
method that would detect the absence of center of symmetry in crystals obtained from
racemic mixtures will be able in most cases to detect conglomerates. To this purpose,
nonlinear optics can be very useful. In the next paragraphs are presented the theory of
second harmonic generation in crystals and described the principle of conglomerate
prescreening.
4.2.1 Nonlinear optics and second harmonic generation (SHG)
In a given medium, the propagation of light is mainly driven by the dielectric properties and
the response to electromagnetic fields. The application of an electromagnetic field (light) to a
molecule modifies the shape of the electronic cloud and consequently creates an induced
electric dipole moment.
Recrystallization
414
The term polarization (P, vectorial dipole moment per unit volume) is used to describe this
phenomenon on a macroscopic scale:
P = c
0
_
(1)
E
(2)
with E, the applied electric field, c
0
is the vacuum permittivity and _
(1)
the linear (first order)
susceptibility of the material.
The linear susceptibility is a 2
nd
rank tensor related to the permittivity c and the refractive
index n of the material:
c = n = c
0
(1 + _
(1)
)
(3)
For week electric fields, the polarization varies linearly with E. However, for high light
intensities (typically greater than 1 MW/cm), the polarization becomes a nonlinear function
of the applied electric field. It can be expressed as a power series expansion of the
macroscopic field:
P = c
0
(_
(1)
E + _
(2)
E
2
+ _
(3)
E
3
+ ) = c
0
_
(1)
E + P
NL
(4)
with _
(2)
and _
(3)
the second and the third order susceptibility tensors respectively and P
NL
,
the nonlinear polarization. This equation implies that during the propagation of light at a
frequency e, nonlinear components of the polarization at frequencies 2e and 3e arise and
harmonics of the original optical field at 2e and 3e are generated . However, with
increasing order, a rapid decrease in the susceptibility coefficients is observed. As a
consequence, the nonlinear polarization P
NL
can be approximated by the quadratic term c
0
_
(2)
E
2
for moderate energies. Considering the usual expression of the electromagnetic field
E = | | c . c ) t ( j exp
2
1
+ e r k. E
0
(5)
with
0
E , the amplitude and k the wave vector and c.c, the complex conjugate of the formula.
the polarization then becomes equal to:
P = c
0
_
(1)
E + c
0
_
(2)
E
2
(6)
| |
| |
e _ c +
e _ c +
_ c
=
) t ( j 2 exp
2
1
) t ( j exp
2
1
2 ) 2 (
0
) 1 (
0
0
) 2 (
0
r k. E
r k. E
E
P
0
0
2
(7)
Finally, the net polarization is composed of three components: a continuous one
corresponding to the phenomenon of optical rectification, a component at frequency e
(optical polarizability) and a second harmonic term (frequency 2e) corresponding to the
phenomenon known as SHG (second harmonic generation). In SHG, a fundamental wave of
amplitude E
e
, angular frequency e (wavelength ) and wave vector k
e
passing through a
crystal generates a second harmonic wave of amplitude E
2e
, angular frequency 2e
(wavelength /2) and wave vector k
2e
.
Recrystallization of Enantiomers from Conglomerates
415
I
e
I
2e
Nonlinear
material
SHG
I
e
I
e
I
2e
Nonlinear
material
SHG
I
e
Fig. 9. Second harmonic generation
With distance traveled through the crystal, the second harmonic amplitude varies. The
energy generated at frequency 2e can be obtained by solving the propagation equation:
2
2
0
2
2
0 0
2
t t c
c
=
c
c
c V
P E
E
(8)
with
0
, the permeability of free space.
The average intensity of an electromagnetic wave of amplitude E is:
I =
c 2
.
0
*
c
E E
(9)
with c, the velocity of light. Assuming that the waves are traveling in the z direction and
that the conversion efficiency is low (so amplitude of the fundamental is almost
unchanged), the second harmonic intensity after a distance through the crystal is
(Armstrong, 1962):
( )
2
2
2
2
2 3
0
2
2
) 2 ( 2
2
I
2
k
2
k
sin
n n c 2
) ( I
e
e e
e
|
.
|
\
| A
|
.
|
\
| A
c
_ e
=
(10)
with :
n
e
, the refractive index of the crystal at angular frequency e
n
2e
, the refractive index of the crystal at angular frequency 2e
k A , the phase mismatch: ) n n (
c
2
k 2 k k
2 2 e e e e
e
= = A ;
I
2e
depends on several parameters and among them on the value of k A . The case of Ak=0
corresponds to materials called phase-matchable materials for which the longer the
distance traveled inside the crystal, the greater the SHG intensity (i.e. large particles will
generate a better SHG signal). Phase-matchable materials such as potassium diphosphate
(KDP) are used to double high power lasers or as a nonlinear standard. But most materials
are non phase-matchable (ie. k A = 0) with as a consequence an intensity value oscillating
with . I
2e
will reach a maximum only for the discrete values of given by:
Recrystallization
416
) n n ( 4 k
2
c
e e
=
A
t
=
(11)
c
is called the coherence length.
This implies that the SHG intensity will be optimized for particles of size equal to
c
or equal
to an odd multiple of
c
. The consequence of the coherence length will be discussed later.
I
2e
depends also on _
(2)
, the second order nonlinear susceptibility, and this is the main point
for the detection of conglomerates. 27 numbers constitute the electro-optic components of
_
(2)
which is
a 3
rd
rank tensor. However, the number of independent coefficients can be
reduced to 10 if the absorption of the material is negligible at e and 2e. In this case, the
tensor is invariant by circular permutation of its three indices.
Moreover, the number of independent coefficients non equal to zero can be determined by
tacking into account the symmetry elements of the 32 crystallographic classes. In particular,
all the components of _
(2)
are
null for centrosymmetric structures. Thus, in crystals with a
centre of inversion, all the components of _
(2)
tensor are zero. Consequently, these types of
crystals can not exhibit a SHG signal. According to the Kleinman symmetry rules
(Kleinman, 1962) three chiral point groups (NC) present also a null _
(2)
(422, 622 and 432).
However, Kleinman symmetry is not always applicable. These different possibilities are
summarized in figure 10.
As a result, the absence of a center of symmetry can be determined via the detection of a SHG
signal. Therefore, this test was chosen as a pre-screening method for spotting conglomerates.
: centrosymmetric point groups (no SHG activity)
0 : non centrosymmetric point groups (SHG activity)
: no SHG activity if Kleinman symmetry is applicable
Fig. 10. SHG activity among the 32 crystallographic classes
Recrystallization of Enantiomers from Conglomerates
417
4.2.2 Experimental set-up
The experimental set up proposed by Kurtz and Perry (Kurtz et al, 1968) gives quick
information about the SHG activity of powder samples. Because the majority of the
pharmaceutical crystalline samples are available in the form of powders, this set-up was
chosen for the pre-detection of conglomerates.
Figure 11, shows the experimental set-up used for the SHG test. The laser is a Nd:YAG Q-
switched laser operating at 1.06 m. It delivers 360 mJ pulses of 5 ns duration with a
repetition rate of 10 Hz. An energy adjustment device is made up of two polarizers (P) and a
half-wave plate (/2).
Optical
fiber
Spectrometer
x
y
z Motorized stage
(Automated 18-wells)
Pulsed laser
Nd: Yag (1.06 m)
5 ns
10 Hz
~ 60 mJ/pulse
Intensity
regulation
Optical
fiber
Spectrometer
x
y
z Motorized stage
(Automated 18-wells)
Pulsed laser
Nd: Yag (1.06 m)
5 ns
10 Hz
~ 60 mJ/pulse
Intensity
regulation
Fig. 11. Experimental Set-up
It allows the incident energy to vary from 0 to circa 200 mJ per pulse. A RG1000 filter is
situated after the energy adjustment device to removes light from laser flash lamps. The
samples (18-wells plate filled with the various powders) are placed on a motorized stage
and irradiated with 60 mJ pulses (beam diameter of 4 mm). The signal generated by the
sample (diffused light) is collected into an optical fiber (500 m of core diameter) and
directed onto the entrance slit of a spectrometer. A boxcar integrator allowed an average
spectrum (spectral range 250-700 nm with a resolution of 0.2 nm) to be recorded over 1
second (10 pulses).
4.2.3 Accuracy of the SHG method
Because of the low occurrence of conglomerates, it would be extremely detrimental to fail to
spot one. Conversely, it is not awkward to have a limited number of false positive
responses. The diagram presented on figure 12 gathers the various situations that can be
encountered during the SHG prescreening. It can be used to evaluate the consistency of the
SHG response observed for a given sample.
Recrystallization
418
(a) Signal detected at /2
Non
centrosymmetric
crystal
Conglomerates
Racemic compound
crystallizing
in non centrosymmetric
space groups 2
Centrosymmetric
crystal
with emission of
spurious light
or
1
Confirmation required by means
of XRPD, spectroscopy,
(b) No signal detected at /2
Centrosymmetric
crystal
Racemic
compound
or
Non
centrosymmetric
crystal
with SHG effect
to low to be
detected
3
4
Racemic compounds
or
conglomerates
from space groups
422, 622, 432
verifying Kleinman
symmetry rules
Absorption
Low non linear
coefficients
Decomposition of the
sample
Inappropriate size
distribution of the powder
Fig. 12. Decision diagram for SHG test
Panel (a) of the diagram corresponds to the samples for which a signal has been observed. It
is necessary to distinguish between two cases: (1) the observation of a non SHG signal
resulting from optical phenomena such as two-photon fluorescence (TPF) or other
photoluminescent processes and (2) the observation of a real SHG signal resulting from
racemic compounds crystallizing in non-centrosymmetric space groups.
False positive response arising from case 1 can be simply encountered. Indeed, TPF signals
present a spectral bandwidth much broader than that of the SHG signal (cf. figure 13).
A base line correction around the SHG wavelength allows the distinction between light
resulting from the SHG process itself (signal of interest) and spurious light. This simple
procedure strongly limits the occurrence of false positive responses.
500 520 540
Wavelength (nm)
S
i
g
n
a
l
(
a
.
u
)
Effective
SHG signal
532 nm
Pure SHG
500 520 540
Wavelength (nm)
S
i
g
n
a
l
(
a
.
u
)
Effective
SHG signal
532 nm
Pure SHG
500 520 540 560
Wavelength (nm)
S
i
g
n
a
l
(
a
.
u
)
No SHG
500 520 540 560
Wavelength (nm)
S
i
g
n
a
l
(
a
.
u
)
Effective
SHG signal
532 nm
non SHG signal
SHG with spurious light
500 520 540 560
Wavelength (nm)
S
i
g
n
a
l
(
a
.
u
)
Effective
SHG signal
532 nm
non SHG signal
SHG with spurious light
Fig. 13. Removal of false positive responses
As mentioned in table 1, it is possible for a racemic mixture to crystallize in a non-
centrosymmetric space group. These types of crystals generate a SHG response and can also
be considered as false positive. They constitute case 2. It is also worth mentioning that
slightly disordered racemic structures and solid solutions can fall in this category. The
influence of the degree of disorder (or pseudo-symmetry) of the structure is difficult to
Recrystallization of Enantiomers from Conglomerates
419
establish so as the consequences on the _
(2)
coefficients. However, the detection of partial
solid solutions can be of interest since some of them can be potentially adequate for
preferential crystallization (Wermester et al, 2007).
Panel (b) of the diagram concerns samples that do not present any signal at /2. The critical
situation of non detected conglomerates is considered in case 3. The main reasons of the no-
detection are an SHG effect too weak to be detected (due to for example to low non linear
coefficients consequence of a low hyperpolarizability of the molecule) or an inappropriate
crystal size distribution. The second harmonic signal generated by a crystalline powder is the
sum of the contribution of each individual particle (electromagnetic fields are uncorrelated).
Because the fundamental beam passes through a large number of particles for which a random
orientation is assumed, the intensity of the SHG beam is not easy to optimize and depends on
the particle size (especially in non phase matchable materials). To preserve the reliability of the
method, too fine particles as those present in submicronic or nanocrystalline powders should
be avoided (coherence length is in order of magnitude of several microns for most materials).
To limit the number of undetected conglomerates, it is also necessary for the experimental set-
up to present a detection level better than 1/100 of the SHG signal generated by the quartz
powder (standard non phase-matchable material for SHG measurements mean diameter 50
m). This condition is in most cases considered as sufficiently constraining. To be exhaustive,
let us mentioned that a non detectable (or a decreasing) signal can also be the consequence of
an absorption in the sample at the wavelength of irradiation () or at the wavelength of the re-
emitted radiation (/2). This absorption can generate a decomposition of the sample when
exposed to the laser beam. The use of an alternative laser source must then be envisaged.
The last case (Case 4) concerns chiral crystals associated to point groups 422, 622 and 432.
These should be SHG inactive due to the application of Kleinman permutation rules.
However, experience refutes the general applicability of these rules. Recently, SHG activity
has been observed in N-acetyl-methylbenzylamine which crystallizes as a conglomerate in
space group P4
1
2
1
2.
Considering all these elements, spotting a SHG active substance can not guaranty the
existence of a conglomerate. Only a complementary study of the samples by conventional
methods can conclude on the conglomerate nature or not. That is why this method is
proposed as a prescreening technique only. However, SHG presents numerous benefits
listed below:
- Only a small quantity of the racemic mixture is necessary (ca. 20 mg typically).
- There is at this first stage of pre-screening no need for comparison between results of
SHG tests on the racemic mixture and the pure enantiomer. The screening can be
undertaken even when the pure enantiomer is not available and thus be carried out at
an early stage of the development of the molecule.
- The response is instantaneously delivered; it is thus conceived to be a true high
throughput pre-screening method which allows on a short period of time to test
numerous derivatives.
- It is a priori a non destructive method.
- It is cheap and can be fully automated: a high throughput device including motorized
sample holders and a computer assisted treatment of the spectra should allow in a
realistic way to select the good candidates with a 50% probability or more.
Recrystallization
420
4.3 Rationalization of conglomerate detection
When a positive SHG signal is obtained, the conglomerate nature of the compound has to be
confirmed via XRPD. However, it is worth mentioning that some bias in the conglomerate
detection can be introduced at this stage if a strict control of the crystallization conditions is
not applied. To avoid misinterpretation of the results, the diagram of figure 14 is proposed
for an optimization of conglomerate detection (Gonella, 2011).
Once the SHG prescreening has led to suspect the existence of a conglomerate (step 1
completed), the next step consists in comparing the XRPD patterns of the racemic mixture
and the corresponding enantiomer. At this stage it is imperative to apply the same
crystallization conditions for the two samples to avoid discrepancies due to polymorphism
or solvation /desolvation. It concerns in particular the nature of the counter-ion or cocrystal
former, the stoichiometry, the temperature (during the crystallization but also during the
SHG detection), the nature of the solvent, etc.
Once both compounds are available (step 2), a precise comparison of the XRPD patterns can
be pursued.
SHG screening
of racemic
mixtures (possibly in
its mother liquor)
SHG
signal +
Comparison of XRPD
patterns of the
racemic mixture
and the enantiomer
Crystallization
of the corresponding
enantiomer
!
Crystallization
conditions must be
identical for
racemic mixture
and pure
enantiomer
No signal
Test more derivatives
!
Check the
cristallinity
Perfect match
Good candidate
for being
a conglomerate
Conglomerate
with partial solid solution
OR
Complete solid solution
Similar with
peaks shifts
PC Tests
Different
Possibility of racemic
non centrosymmetric
crystal
OR
Variations
of the experimental
conditions
(analysis and/or
crystallization)
SHG screening
of racemic
mixtures (possibly in
its mother liquor)
SHG
signal +
Comparison of XRPD
patterns of the
racemic mixture
and the enantiomer
Crystallization
of the corresponding
enantiomer
!
Crystallization
conditions must be
identical for
racemic mixture
and pure
enantiomer
No signal
Test more derivatives
!
Check the
cristallinity
Perfect match
Good candidate
for being
a conglomerate
Conglomerate
with partial solid solution
OR
Complete solid solution
Similar with
peaks shifts
PC Tests
Different
Possibility of racemic
non centrosymmetric
crystal
OR
Variations
of the experimental
conditions
(analysis and/or
crystallization)
Fig. 14. Workflow for conglomerate detection
Three different cases can be encountered on step 3 : perfect match, similar diffractograms
with some peak shifts or different diffractograms). On Figure 15 are presented the XRPD
patterns obtained for three different compounds. The upper traces correspond to the
racemic mixture (1a, 2a, 3a) and the lower ones to the pure enantiomer (1b, 2b, 3b).
Diffractograms 1a and 1b perfectly match which permits to conclude that compound 1 is
with a high level of confidence a conglomerate. PC tests can be undertaken. Diffractograms
2a and 2b present numerous differences. Compound 2 is unlikely to be a conglomerate
Recrystallization of Enantiomers from Conglomerates
421
Fig. 15. Comparison of X-Ray patterns for the detection of conglomerate
Recrystallization
422
forming system. However, it is worth mentioning that interesting conglomerates may exist
only in equilibrium with their mother liquor (because of one or several of the following
reasons: non congruent solubility, solvates with an efflorescent character, solvates which are
hydrolyzed by moistures, CO
2
sensitive solid, etc). So it is of primary importance to check if
the conditions used during the SHG test are identical to those used for XRPD. Indeed, due to
the quickness of the SHG test, an efflorescent solvate could be detected as a conglomerate.
Its desolvation prior or concomitantly to the XRPD analysis could then lead to contradictory
results. To overcome these problems, SHG and XRPD tests should be run under strictly
identical conditions of crystallization.
Diffractograms 3a and 3b are extensively similar but show slight shifts in 2u positions of
some peaks. This couple of derivatives deserves more investigations. Nevertheless, it is
likely that this system exhibits at least partial solid solutions.
5. Conclusion
The objective of this chapter was a better understanding of chiral discrimination in the solid
state and of the mechanisms involves during recrystallization of enantiomers.
We described two methods of enantiopurification and underlined the benefits of
preferential crystallization. Because the detection of a conglomerate is an essential step of
this process, we detailed a recently developed technique of prescreening of conglomerates.
We finally proposed a workflow to follow in order to optimize the detection of
conglomerates and avoid bias that can be induced by different crystallization parameters.
6. References
Armstrong, J. A., Bloembergen, N., Ducuing, J., & Pershan, P. S. (1962). Interactions between
light waves in a nonlinear dielectric. Physical Review, Vol. 127, No. 6, (September
1962), pp. 19181939
Belsky, V. K. & Zorkii, P. M. (1977), Distribution of organic homomolecular crystals by chiral
types and structural classes, Acta Crystallographica Section A, Vol. 33, No. 6,
(November 1977), pp. 10041006, ISSN 1600-5724
Brock, C. P., Schweizer, W. B., & Dunitz, J. D. (1991), On the validity of wallach's rule: on the
density and stability of racemic crystals compared with their chiral counterparts,
Journal of American Chemical Society, Vol. 113, No. 26, (December 1991), pp. 9811
9820, ISSN 0002-7863
Claborn, K., Puklin-Faucher, E., Kurimoto, M., Kaminsky, W., & Kahr, B. (2003), Circular
dichroism imaging microscopy: application to enantiomorphous twinning
in biaxial crystals of 1,8-dihydroxyanthraquinone, Journal of the American Chemical
Society, Vol. 125, No. 48, (December 2003), pp. 1482514831, ISSN-0002-7863.
Collet, A. (1999), Separation and purification of enantiomer by crystallization methods,
Enantiomer, Vol. 4, No. , (1999), ISSN
Coquerel, G., Petit, M.-N., & Bouaziz (1995). Method of resolution of two enantiomers by
crystallisation. PCT Patent WO 95/08522.
Coquerel, G.; Petit, M.-N.; Bouaziz, R.(1994) Procd de ddoublement (AS3PC) de deux
antipodes optiques par entranement polythermique programm et auto-
ensemenc. PCT N94/01.107, 22/09/1994, 1994.
Recrystallization of Enantiomers from Conglomerates
423
Coquerel, G. (2000), Review on the heterogeneous equilibria between condensed phases in
binary systems of enantiomers, Enantiomer, Vol. 5, No. 5, pp. 481-498, (Mai 2000),
ISNN 1024-2430
Coquerel, G. (2007). Preferential Crystallization, In : Novel optical resolution technologies -
Topics in Current Chemistry, Sakai, Kenichi et al (Eds), pp. 151, Springer, ISBN 978-
3-540-46317-7, Berlin.
Dalhus, B. & Grbitz, C. H. (2000), Non-centrosymmetric racemates: space-group
frequencies and conformational similarities between crystallographically
independent molecules, Acta Crystallographica Section B, Vol. 56, No. 4, (August
2000), pp. 715719, ISSN 1600-5740
Flack, H. (2003), Chiral and achiral crystal structures, Helvetica Chimica Acta, Vol.86, No.4,
(January 2003), pp. 905921, ISSN 1522-2675
Flack, H. D. & Bernardinelli, G. (2003), The mirror of galadriel: looking at chiral and achiral
crystal structures, Crystal Engineering, Vol. 6, No. 4, (December 2003) pp. 213223,
ISSN 1463-0184
Galland, A., Dupray, V., Berton, B., Morin-Grognet, S., Sanselme, M., Atmani, H., &
Coquerel, G. (2009), Spotting conglomerates by second harmonic generation.
Crystal Growth & Design, Vol. 9, No. 6, (May 2009), pp. 27132718, ISSN 1528-7483
Gernez, D. (1866) Sparation des tartrates droits et des tartrates gauches laide de solutions
satures, Compte-rendus de lAcadmie des Sciences, Vol. 63, (July-December 1866), pp.
843-888
Gonella, S., Mahieux, J., Sanselme, M., & Coquerel, G. (2011). Spotting a conglomerate is just
halfway to achieving a preparative resolution by preferential crystallization.
Organic Process Research & Development, In press, ISSN 1083-6160.
Helmreich, M.; Niesert, C.-P.; Cravo, D.; Coquerel, G.; Levilain, G.; Wacharine-Antar, S.;
Cardinal, P, (2010), Process for enantiomeric separation of racemic dihydro-1, 3, 5
triazines via preferential crystallization. PCT patent WO2010109015 (A1), 2010-09-30
Jacques, J., Collet, A., & Wilen, S. (1994). Enantiomers, Racemates and Resolutions (3rd
Edition), Kriger Pub. Co., ISBN 0894-648764, Malabar Florida, USA.
Levilain, G., Rougeot, C., Guillen, F., Plaquevent, J-C., Coquerel, G. (2009) Attrition
enhanced preferential crystallization combined with racemization leading
toredissolution of the antipode nuclei, Tetrahedron: Asymmetry, Vol. 20, No. 24,
(December 2009), pp. 27692771, ISSN 0957-4166
Levilain, G., Coquerel, G. (2010) Pitfalls and rewards of preferential crystallization,
CrystEngComm, Vol. 12, No. 7, (May 2010), pp. 1983-1992, ISSN 1466-8033
Jungfleish, M. E. (1882) , Journal of Pharmaceutical Chemistry, Vol. 5, pp. 346.
Kaptein, B., Elsenberg, H., Grimbergen, R.F.P., Broxterman, Q.B, Hulshof, L.A., Pouwer,
K.L. & Vries, T.R. (2000) Dutch resolution of racemic 4-hydroxy- and 4-
fluorophenylglycine with mixtures of phenylglycine and (+)-10-camphorsulfonic
acid, Tetrahedron: Asymmetry, Vol. 11, No. 6, (April 2000), pp. 1343-1351, ISSN 0957-
4166
Kleinman, D. A. (1962). Theory of second harmonic generation of light. Physical Review, Vol.
128, No. 4, ( 1962) pp. 17611775.
Kostyanovsky, R. G., Kostyanovsky, V. R., & Kadorkina, G. K. (2009) The enigma of a ()-
tartaric acid-urea cocrystal, Mendeleev Communications, Vol. 19, No.1, (2009 ), pp.
17- 18, ISSN 0959-9436
Recrystallization
424
Kuroda, R. & Honma, T. (2000) Cd spectra of solid-state samples, Chirality, Vol. 12, No. 4,
(April 2000), pp. 269277, ISSN 0899-0042
Kurtz, S. & Perry, T. (1968) A powder technique for the evaluation of nonlinear optical
materials, Journal of Applied physics, Vol.39, No. 8, (July 1968), pp.37983813, ISSN
0021-8979.
Lorenz, H., Perlberg, A., Sapoundjiev, D., Elsner, M. P., & Seidel-Morgenstern, A. (2006),
Crystallization of enantiomers, Chemical Engineering and Processing: Process
Intensification, Vol. 45, No. 10, (April 2006), pp. 863 873, ISSN 0255-2701
Marchand, P., Lefebvre, L., Querniard, F., Cardinael, P., Perez, G., Counioux, J-J & Coquerel,
G. (2004), Diastereomeric resolution rationalized by phase diagrams under the
actual conditions of the experimental process, Tetrahedron: Asymmetry, Vol.15,
No.16, (August 2004), pp. 2455-2465, ISSN 0957-4166
Marckwald, W., (1896) Ueber ein bequemes Verfahren zur Gewinnung der Linksweinsure,
Berichte der deutschen chemischen Gesellschaft, Vol.29, No.1, (April 1896), pp.42
43
Ndzi, E., Cardinael, P., Schoofs, A. R., & Coquerel, G. (1997), An efficient access to the
enantiomers of [alpha]-methyl-4-carboxyphenylglycine via a hydantoin route using
a practical variant of preferential crystallization as3pc (auto seeded programmed
polythermic preferential crystallization), Tetrahedron: Asymmetry, Vol. 8, No. 17,
(September 1997), pp. 29132920, ISSN 0957-4166
Kocher-Becker, U. , Kocher, W. & Ockenfels, H. (1982), Teratogenic activity of a hydrolytic
thalidomide metabolite in mice, Naturwissenschaften, Vol. 69, No. 4, (Avril 1982),
pp. 191-192, ISSN 0028-1042
Pasteur, L. (1853), Transformation des acides tartriques en acide racmique - Dcouverte de
l'acide tartrique inactif. Nouvelle mthode de sparation de l'acide racmique en
acides tartriques droit et gauche. Compte-rendus de lAcadmie des Sciences, Vol. 37,
(Juillet 1853), pp.162-166
Pope, W.J. & Peachey, S.J. (1899), The application of powerful optically active acids to the
resolution of externally compensated basic substances. Resolution of
tetrahydroquinaldine, Journal of Chemical Society Transactions, Vol. 75, (January
1899), pp. 1066-1093, ISNN 0368-1645
Renou, L., Morelli, T., Coste, S., Petit, M.-N., Berton, B., Malandain, J.-J. & Coquerel, G.,
(2007), Chiral discrimination at the solid state of the methyl 2-
(diphenylmethylsulfinil)acetate, Crystal Growth & Design, Vol. 7, No. 9, (August
2007), pp.1599-1607, ISSN 1528-7483
Stinson, S.C. (2000). , Chiral Drugs, Chemical and Engeniering news, Vol. 78, No. 43, pp. 55-78
(October 2000), ISSN 0009-2347
Subramanian, G., (2006). Chiral Separation Techniques: A Practical Approach (3
rd
18
th
October), Wiley-VCH Verlag GmbH, ISBN: 978-3-527-31509-3, Weinheim, Germany.
Vries, T., Wynberg, H.,Van Echten, E.A, Koek, J.A, Ten Hoeve, W.A., Kellogg, R.M., Broxterman,
Q.B., Minnaard, A.B, Kaptein, B.B, Van Der Sluis, S., Hulshof, L. & Kooistra, J. (1998),
The family approach to the resolution of racemates, Angewandte Chemie - International
Edition, Vol. 37, No 17, (September 1998), pp. 2349-2354, ISSN 1433-7851
Wermester, N., Aubin, E., Pauchet, M., Coste, S., & Coquerel, G. (2007). Preferential
crystallization in an unusual case of conglomerate with partial solid solutions.
Tetrahedron: Asymmetry, Vol. 18, No. 7, (2007), pp. 821831, ISSN 0957-4166
18
Recrystallization of Drugs: Significance on
Pharmaceutical Processing
Yousef Javadzadeh, Sanaz Hamedeyazdan and Solmaz Asnaashari
Biotechnology Research Center and Faculty of Pharmacy,
Tabriz University of Medical Sciences
Iran
1. Introduction
Not surprisingly, the wide range of effective medicinal agents available today is one of the
greatest scientific achievements. Regardless of the advancements in effectiveness and safety
of the medicines embedded in dosage forms, the pharmaceutical concept of the latter is
growing to be ever more eminent (Adibkia et al., 2011). Following on from recent
advancements, in a time of increased considerations to the level of sophistication in
designing pharmaceutical dosage forms keeping pace with advances in drug discovery
methods, it seems as important as ever to study the physicochemical properties of active
pharmaceutical ingredients, prerequisite for a successful product formulation.
As far as we know, the molecular structure of any drug compound typically defines all of its
physical, chemical and biological actions. Owing to the fact that a certain kind of drug might
be offered in a variety of solid forms, including polymorphs, solvates, hydrates, salts, co-
crystals and amorphous solids, the choice and design of the ideal solid-state chemistry of the
pharmaceutical solid form would be critically important to a superior drug development.
Accordingly, drug crystals could be modified in different ways including recrystallization,
which would affect the physical and physicochemical properties such as melting point,
solubility, true density, drug release profile, flowability and tabletability of the pharmaceutical
dosage forms (Harbury, 1947; JamaliMitchell, 1973; Jozwiakowski et al., 1996).
Recrystallization is a simple and inexpensive method for scaling up the drug developments
to a commercial level. Significant advances in the different pharmaceutical dosage form
technologies renders drug recrystallization as a green technique due to the savings of costs,
time, energy and less machinery as well as fewer personnel. Recrystallization is one aspect
of precipitation obtained through a variation of the solubility conditions and the amount of
dissolved solute in an increased temperature. In general, production of another crystalline
from of a drug and also purification are the two major sets of applications for drug
recrystallization processes. Briefly, in a drug recrystallization process, a hot saturated
solution of the drug is prepared with only enough solvent to dissolve it at the boiling point
of the solution. Once the solution is cooled the purified drug component or a new crystal
form of the drug separates as a result of the lower solubility of that crystalline form of a
drug in the respective solvent at lower temperatures.
Recrystallization
426
Since impurities are present in fairly small amounts of drug solutions they do not crystallize in
recrystallization and they are ready to separate from the formed drug crystal (Tiwary, 2001).
For instance, in the case of natural medicinal compounds obtained from natural sources which
almost always contain impurities, in order to obtain a pure drug, usual major steps in the
recrystallization process have been schematically demonstrated in figure 1. Purifying a sample
drug of compound X which is contaminated by a small amount of compound Y, would be
established with an appropriate hot solvent in which all of compound Y is soluble at room
temperature and the impurities will stay insoluble in and pass through filter paper, leaving
only pure drug crystals behind, as has been shown in figure 1.
Fig. 1. Schematic protocol of drug recrystallization for further purification.
It is worth mentioning that in any recrystallization technique some drug loss is inevitable
and the total recovery would be less than 100%, seeing that even at the lower temperatures
the target drug has some finite solubility in the solvent and is lost subsequently when
solvent and soluble impurities are removed. Moreover, selection of the right solvent in the
recrystallization seems to be one of the crucial features of the process and is made on a case-
by-case basis (MirmehrabiRohani, 2005; ChenTrout, 2008). This is because of solubility
variation of different drug compounds in different solvents so that a certain drug not only
should have the highest solubility in the solvent of choice at its boiling point, but also it
should show a markedly diminished solubility at lower temperatures of the same solvent.
Although recrystallization is a very common technique used to purify drugs, it has a basic
limitation for the compounds that are mostly pure and other techniques of separations are of
use for the drug mixtures containing several major components which could not be purified
by recrystallization methods.
Ever since, the leading physicochemical properties of a unique form of a drug could
seriously influence the bioavailability, manufacturability purification, stability, solubility,
and other characteristics, identifying these potential liabilities allows us to predict, control
Recrystallization of Drugs: Significance on Pharmaceutical Processing
427
and avoid any complexities that may arise during drug development stages (Krishnaiah,
2010). This would be beneficial in preventing the drug development efforts to costly late
stage product failures throughout the manufacture and storage periods too. However, in
order to fully control the crystallization process, the link between a particular solid form of a
drug structure and its functional physicochemical properties, still challenges to be better
established to facilitate the suitable drug production. A large and growing body of
literatures has been published on recrystallization techniques and physicochemical
properties of a drug. Herein, we gathered some of the related reports of drug
recrystallization to have an overlook on the crystal habit of a drug on some basic physical
properties of pharmaceutical dosage forms signifying how these factors are interrelated.
2. Impact of crystal habit on pharmaceutical processing
Drug discovery and characterization relies on the nature of the target molecule and the
relative physicochemical properties of drugs. Identifying all relevant crystal habit of a drug
which is an important variable in pharmaceutical manufacturing at the development phase
from research to commercialization is of substantial value. Due to the different crystal form
variations of some basic physical properties like, solubility, dissolution rate, melting
behavior, and certain micromeritic properties or performance characteristics, e.g. tablet
compressibility, mechanical strength, powder flow provide alternatives to select a form that
presents the suitable balance of critical properties for development into the drug product.
Establishing such modification information at an early stage of drug development process
lessens the risk of process alterations given form changes and brings in the opportunity to
attain more comprehensive rational property coverage.
The merit of changes in crystal surface form and habit of drug powders by recrystallization
method is much more realized when there is an essential to diminish variations in raw
material characteristics, to certify reproducibility of results during drug preformulation, and
also to judge fairly about the cause of poor performance of a dosage form. Besides, the
changes in crystal habit of a drug going together with or without polymorphic
transformation at some point in processing storage could account for serious implications of
physical stability in dosage forms. Thus, it seems underlying to have a deeper insight to the
crystal structures and control the solid-state chemistry of drug substances to design a more
systematic and intellectual pharmaceutical dosage forms.
In a survey carried out by Sinclair et al. ibipinabant a potent and highly selective
cannabinoid receptor antagonist was evaluated for its solid-state physical stability and
recrystallization kinetics in tablet dosage forms using fourier transform raman spectroscopy.
The findings of the study showed that exposure to moisture had notable influence on the
crystallinity of amorphous ibipinabant. The recrystallization kinetics measurements
revealed a two-step process with an induction period (nucleation) followed by rod-like
crystal growth by application of the JohnsonMehlAvrami kinetic model. On the whole
their method provided reliable and highly accurate predictive crystallinity assessments after
exposure to a variety of stability storage conditions for ibipinabant (Sinclair et al., 2011).
Recently, Dahlberg et al. analyzed the stability of the amorphous drug, flutamide, by a
combination of localized nuclear magnetic resonance (NMR) spectroscopic and NMR
imaging techniques. Owing to the fact that, NMR relaxation is sensitive to both the
Recrystallization
428
crystalline and amorphous state and the size of the drug substance, it allows for an in situ
monitoring of the state of the drug during tablet disintegration and dissolution periods.
With regard to the results of the NMR experiments, recrystallization was believed to be
related to its enabling factors such as local hydration level and local mobility of the polymer
matrix. Eventually, it was verified that the primarily amorphous flutamide may recrystallize
either by nanoparticle coalescence or by ripening of crystalline particles (Dahlberg et al.,
2011).
The solid-state properties of sulfathiazole and chlorpropamide were modified through
recrystallization using supercritical antisolvent process by Yeo et al. They confirmed that the
operating conditions of the system such as carbon dioxide injection rate, type of solvent, and
temperature significantly had an effect on the physical characteristics of the resulting
crystals. Considering the results of the study, drug crystals processed with supercritical
system exhibited more ordered appearances with clean surfaces and sharp angles compared
with the unprocessed particles where crystal habit changed from tabular to acicular when
the carbon dioxide injection rate increased. Photomicrographs of sulfathiazole crystals with
methanol as a solvent, confirmed a needle-like acicular and a tabular crystal habits in rapid
and slow injections, respectively. Whereas, in the case of chlorpropamide, processed drug
particles in the rapid injection experiment exhibited columnar habit in a regular shape,
while relatively large crystals with sharp angles were observed in the slow injection mode
when acetone was used as the solvent. Overall experimental observations suggested that the
supercritical antisolvent process could provide favorable environment for the solid growth
of a single type of crystalline drug, minimizing the conditions for growth-related
imperfections (Yeo et al., 2003).
According to the fact that thermal analysis has been frequently used to identify crystal forms
of drugs and in the course of thermal analysis, crystal transformation is often observed as
well as melting and decomposition, Suzuki et al. studied mechanisms of thermal crystal
transformation through melting and recrystallization. They characterized two anhydrates
(-from and -from) and two hydrates (hemihydrate and monohydrate) forms of a novel
fluoroquinolone antibiotic, sitafloxacin, in addition to sesquihydrate which is used in the
marketed drug products. The results of crystal structural that were characterized by infrared
spectroscopy, X-ray powder diffractometry and thermal analysis revealed quinolone rings
of sitafloxacin had distorted planar structure and quinolone ring of the drug in -form and
monohydrate hold opposite torsion to those in -form and sesquihydrate. These kinds of
thermal analysis are often recommended as a routine tool for quality control of thermal
dehydration and subsequent crystal conversion of drugs (Suzuki et al., 2010).
2.1 Compaction and flowability
Acquiring a clear notion of why certain drug materials are prone to problems during
compaction and dominate the relative constraints to offer a successful compaction and
tableting strategy of pharmaceutical powders would involve an understanding of the
fundamental properties of drug powders. Therefore, it is important to determine the effect
of different physicochemical properties such as particle size, shape, surface area,
polymorphic form, crystal habit, hydrates, and processing conditions on the compaction of
powders. As different crystal habits of a certain drug hold dissimilar planes, they have
differing points in their specific surface and free surface energies. Even so, alternative
Recrystallization of Drugs: Significance on Pharmaceutical Processing
429
recrystallization solvents could develop crystal with defined crystal habit, size and shape as
well as compressibility properties. The nature and amount of these changes count on the
recrystallization conditions including the presence of impurities, type of solvents and
cooling rates. This view is supported by a variety of papers at molecular level developing
the knowledge of solid-state properties such as crystal structure, crystal habit, and
polymorphism influence on the mechanical properties of powders in an attempt to identify
and modify physical properties of bulk solids of drugs (Liebenberg et al., 1999; Maghsoodi
et al., 2007).
Seton et al. evaluated the particle morphology of ibuprofen, an anti-inflammatory drug, by
recrystallization from a range of solvents and investigated the following influence on
compaction properties. The compaction data achieved from properties of the ibuprofen
control and recrystallized samples at different compaction forces and speeds revealed equal
or better tablet strength than the control, whilst ibuprofen recrystallized from 2-ethoxyethyl
acetate exhibited lower levels of elastic energy during compaction. In addition, the
recrystallized ibuprofen samples demonstrated flowablity equivalent quality to the
ibuprofen control, excluding the ibuprofen recrystallized from acetone which showed
excellent flow properties. Generally, the results displayed ibuprofen recrystallization from
various solvents could offer advantages in terms of particle morphology, flowability and
compaction properties (Seton et al., 2010).
In another study an anti-epileptic drug, phenytoin, crystals in the form of free acid, having
distinct types of habits, was modified via different recrystallization conditions and
techniques by Nokhodchi et al. Several sets of experimental conditions for temperature,
solvent evaporation and watering-out techniques were applied for evaluation of the drug
recrystallization in ethanol and acetone solvents. The solid state characteristics and
compaction properties of the crystal habit with factors affecting the resultant crystals were
also evaluated. The physical characteristics of the crystals were investigated using scanning
electron microscopy, X-ray powder diffractometry, FT-IR spectrometry and differential
scanning calorimetry. They confirmed that using watering-out technique as a crystallization
method, produced thin plate crystals, while the crystals obtained by other methods were
needle shape for alcoholic solutions and rhombic for acetone solutions. Although the
crystallization medium had central effect on phenytoin crystal habit modification, altering
crystallization temperature had no effect on crystal habits except a change in size of crystals.
In the case of compaction, the crystals produced from alcohol or acetone showed high
crushing strengths as a result of lower porosity and lower elastic recoveries (Nokhodchi et
al., 2003).
As we know, ascorbic acid crystals are unsuitable for direct tableting due to their poorly
compactible properties, Kawashima et al. designed spherically agglomerated crystals of
ascorbic acid with improved compactibility for direct tableting. They precipitated ascorbic
acid crystals by a solvent change method, followed by their agglomerations with the
emulsion solvent diffusion or spherical agglomeration mechanism, depending on the
solvent combination for crystallization. Considering the results of the study, under static
compression, effectively the proper compact with a sufficient strength was produced. After
all improved micromeritic properties, such as flowability and packability for the spherically
agglomerated crystals were obtained for crystals of ascorbic acid with the spherical
crystallization technique (Kawashima, 2003).
Recrystallization
430
Designing a suitable dosage form with an ideal physicochemical and mechanical property is
an important basic principle of drug delivery systems. As follows, crystal structure, shape,
and size of drug substances have a huge economical and practical effect at all stages of
development from research to commercialization. So, there is a necessity to control the
critical properties of drugs for their readiness and capacity to form a tablet which are
dominating dosage forms in pharmaceutical dosage form manufacturing.
2.2 Solubility, dissolution, and bioavailability of drugs
Nowadays, in pharmaceutical companies drugs with restricted aqueous solubility have
become ever more prevalent and challengeable in the research and development stages.
Slow drug dissolution in biological fluids, insufficient and inconsistent systemic exposure
and subsequent inadequate efficacy in patients, are some of routine challenges to be coped
with during the development of poorly water-soluble drug substances especially when they
are administrated orally. Notable numbers of drugs especially new drug candidates are in a
biopharmaceutical classification of low solubility (BCS Class II and IV) keeping drug
dissolution rate as the limiting factor for the drug absorption and attaining suitable blood-
levels of the drugs (LobenbergAmidon, 2000).
These inadequacies in solubility of clinically
established drug substances in water and in the gastric fluids make problems in drug
dissolution rate and oral bioavailability of drugs, as well (Blagden et al., 2007).
Consequently, there is a basic requisite to deliver such drugs in a way that gives a chance of
sufficient dissolution rate, absorption,and demonstrating suitable clinical efficacy.
Numerous scientific and technological advancements have been made in the research and
development for improving and maximizing dissolution rates of the mentioned types of
drugs. Despite enhancements in solubility and dissolution rate and oral bioavailability of
poorly water-soluble drugs with the customary pharmaceutical technologists, still there are
concerns about the success of those methods in the complexities arise from the specific
physicochemical nature of the drug molecule itself (Krishnaiah S.R., 2010). One of the
thriving trends in enhancing the solubility, dissolution rate and subsequent bioavailability
of poorly soluble drugs is to deal with crystal forms of materials which could potentially be
applicable to a broad range of drugs with different crystalline habits (Yeo, Kim et al., 2003).
On account of many factors such as crystal habit, size and even polymorphic forms of a
drug, dissolution rates would enhance through habit recrystallization. A number of reports
in the literature validated the effects of crystal morphology variation on solubility, in vitro
dissolution rate, holding potentials for improving drug bioavailability (Kawashima et al.,
1986; Carino et al., 2006).
Several studies in this filed have shown that exposure of diverse crystal faces determines
the nature of the wettability and consequent enhancements in dissolution rate of the
drugs with different crystalline shapes (Heng et al., 2006). In 2000, Kobayashi et al.
published a paper in which they presented different dissolution rates for carbamazepine,
where the dihydrate form of the drug in simulated fluids (pH 1.2) had notably slower
dissolution rates than the anhydrous forms (forms I and III). Although the metastable
polymorph (Form III) possessed greatest rates of dissolution at the initial stages,
reductions in dissolution rate at later time points of the profile was achieved due to the
rapid conversion of metastable polymorph (Form III) to the dihydrate. Nevertheless, in
another study carried out by Tian et al. the behavior of carbamazepine and dihydrate
Recrystallization of Drugs: Significance on Pharmaceutical Processing
431
compacts during in vitro dissolution tests various factors were evaluated. Considering the
results, presence of excipients such as polyethylene glycol (PEG) and hydroxyl propyl
methyl cellulose (HPMC) inhibited the conversion of carbamazepine to the hydrated form
following decreased rates of drug dissolution (Tian et al., 2007). Application of different
drug habits in pharmaceutical dosage forms could vary the dissolution rates, as the use of
metastable polymorphs in enhancing drug dissolution rates. They also performed
bioavailability tests in dogs to determine the effects of physicochemical properties of drug
form I, form III and dihydrate on the plasma level of carbamazepine. Similar to the
findings of other dissolution studies for carbamazepine, drug bioavailability that had
been measured was lowest for the dehydrate form. The lower drug bioavailability
established with metastable form was in consistent with the probable conversion of the
drug habit to the dihydrate (Kobayashi et al., 2000).
Intrinsic solubility of three crystals of diclofenac, a nonsteroidal anti-inflammatory drug,
was investigated by Llinas et al. The crystal habits were characterized and detected by
thermo gravimetric analysis, differential scanning calorimetry, and X-ray diffraction. They
recrystallized the anhydrous sodium salt of commercially available diclofenac with ethanol
and precipitated as a hydrated drug that provided consistent results for the intrinsic
solubility. Regarding the broad range of values which have been reported for aqueous
diclofenac solubility in the literature, they claimed their solubility records were at the
smaller end of the range (Llinas et al., 2007).
Perlovich et al. analyzed four new crystal structures of the sulfonamides by X-ray diffraction
experiments and comparative analysis of molecular conformational states and hydrogen
bonds networks by graph set notations in the crystal lattices. They established temperature
dependencies of the solubility in water, n-octanol as well as thermodynamic functions of
solubility and solvation processes for the compounds. According to Perlovich et al.
distinguishing between enthalpy and entropy leads to the insight that the mechanism is
different for the different molecules where it may be of importance for further assessment of
distribution of drug molecules and provide a better understanding of biopharmaceutical
properties of drugs (Perlovich et al., 2008).
Dipyridamole as a critical antiplatelet and peripheral vasodilator drug is known to have
properties of water insolubility and poor bioavailability which are the limitations of its
effectiveness in clinical usage. Adhiyaman et al. characterized dipyridamole crystals
with different types of habits by recrystallization from selected solvents.
Physicochemical characteristics of the crystals were assessed via scanning electron
microscopy, X-ray powder diffractometry, IR spectrometry and differential scanning
calorimetry. The developed crystals of dipyridamole under optimized conditions ensue
in different crystalline habits that significantly improved dissolution rate compared to
original dipyridamole. Recrystallized dipyridamole with benzene and acetonitrile,
produced needle shaped crystals and the ones recrystallizaed with methanol produced
rectangular shaped crystals. Whereas smooth needle shaped crystals were obtained with
the methanolic solution of the drug in the presence of Tween-80, Povidone K30 and PEG-
4000 (AdhiyamanBasu, 2006). Generally, these results were in consistence with the
possibility of controlling and enhancing the drug release properties following by a
probable improvement in bioavailability of drug particles through characterization of
drug crystals.
Recrystallization
432
Recrystallization of phenytoin in ethanol and acetone by Nokhodchi et al. was shown to
produce needle-like and rhombic crystal habits which brought about identical dissolution rates
of crystals obtained from both solvents. Considering the results, nature of recrystallization
solvents in this case had no effect on dissolution profiles. It was suggested that the differences
in dissolution rates for phenytoin was related to the surface area of various crystals with
different shapes (Nokhodchi, Bolourtchian et al., 2003). Wettability and the changes in intrinsic
dissolution rate of doped phenytoin crystals were evaluated by Chow et al. They stated that
the differences in dissolution rates of phenytoin crystalline powders with different
morphology were mainly because of the changes in surface area rather than the improvements
in the wetting of more polar surface moieties. However, they stated that the areas of the
relatively polar faces seem to be valuable determinants of the drug release profiles of doped
phenytoin crystals along with the correlation of the surface tensions (Chow et al., 1995).
Talinolol is a cardioselective beta blocker agent that is known to have different crystal
structures with strongly differing solubilities when pure water, acetate, or phosphate buffers
are employed as dissolution media. Wagner et al. have studied the impact of different
dissolution media controlling the crystal structures of talinolol influencing the dissolution
rate and solubility of the drug. The crystal structures were analyzed by means of light
microscopy, differential scanning calorimetry, and X-ray powder diffraction, detecting the
variations of talinolol crystal structures being the source of incomplete and unpredictable
nature of the drug bioavailability (Wagner et al., 2003).
Carbamazepine, a routinely used drug in the treatment of epilepsy and trigeminal neuralgia,
exists in four polymorphic forms and as a hydrate which could modify the physicochemical
properties of the drug. In our previously published paper we established enhanced
physicomechanical properties of carbamazepine via recrystallization at different pH values.
The resultant habits of carbamazepine crystals varied from flaky or thin plate-like to needle
shape structures which were ascertained using scanning electron microscopy and X-ray
powder diffraction. Considering the results of the in vitro dissolution evaluations of
carbamazepine samples, a higher dissolution rate for carbamazepine crystals were obtained
from media with pH 11 and 1 compared to the original carbamazepine sample. After all, the
carbamazepine particles recrystallized from aqueous solutions with different pH values
revealed superior mechanical properties which were generally in consistence with the similar
studies of drug recrystallization (Grzesiak et al., 2003; Javadzadeh et al., 2009).
In spite of the absolute potential of drug habit modifications in dissolution rate promotion,
far too little attention has been paid to gather detailed documentations of the usage of the
approach in enhancement of systemic drug efficiencies following drug bioavailability in
human subjects or in suitable animal models. So as to affirm drug recrystallization as an
efficient practice in intentionally increasing the bioavailability of poorly soluble drugs,
further investigation in this field is mandatory. Ultimately, we could consider the potential
management of crystal habits of poorly water soluble drugs as an approach for designing
efficient pharmaceutical dosage forms.
3. Polymorphism
Very early on, in 1832 Whler and Liebig reported the first observation of polymorphism
upon cooling a boiling solution of benzamide where needle-shaped crystals would initially
Recrystallization of Drugs: Significance on Pharmaceutical Processing
433
formed followed by a conversion to rhombic crystals upon standing (WhlerLiebig, 1832).
After a longtime history, polymorphism has maintained its innovation for scientists as a
curiosity and an urgent challenge of commercial relevance in manufacturing industries, as
well.
In general, polymorphism is known to be the ability of a compound to crystallize in more
than one distinct crystal structure. Evidently, structures of different crystals would receive
various scopes of the possible ranges of intermolecular interactions like, van der Waals,
ionic, and hydrogen bonds. It would not be far from expectations that the different
polymorphs of the same molecule will have different free energies affecting all the basically
physicochemical properties of the compound, in consequence. Therefore, the crystal
structure of drugs have leading signature on both physical and chemical properties in a way
that the solid-state forms might demonstrate variations in, color, stability, processability,
solubility, dissolution and bioavailability, ranging from the subtle to the severe (Rodriguez-
Spong et al., 2004). Despite the fact that distinct crystal habits have different processing
issues or different chemical stability, these variations usually have direct solutions and the
real impact of crystal forms is the difference in solubility and bioavailability. Figure 2 gives a
fair picture of the probable consequences of a different crystal form in solubility. Decreased
solubility of a certain crystalline form of a drug not only brings about lower bioavailability
but also reduced rates of drug clearances are inevitable which are the inferences to the safety
and efficacy of any drug product.
Fig. 2. Relative solubility of different crystal forms as a function of thermodynamic stability
Normally, producing the most thermodynamically stable polymorphic form of a drug in
the range of interest is compulsory. The pharmaceutical science has been chiefly in charge
for a change in this situation ever since the majority of drugs delivered orally receive rigid
approval for a single crystal form or polymorph. Establishing the most
thermodynamically stable form of a drug relies on obtaining comprehensive information
about the existence and the interrelation of the polymorphs for a given active ingredient.
Different polymorphs of a drug could be prepared by recrystallizing the drug with a
range of solvents under the optimized conditions which has currently been arisen as an
active research province of pharmaceutical science for improving the formulation related
problems of drug molecules.
Recrystallization
434
Accordingly, drug polymorphism investigations have an important role in any
reformulation study since it has an impact on the development potential of a drug molecule,
so as understanding the polymorphic tendencies of a drug molecule thorough
characterization of the observed forms is of prime value.
The first important step in characterizing a polymorph of a drug is to distinguish between
different structures of a molecule by its class. As it has been shown in figure 3, the most
commonly observed forms in drug solids are the polymorphs, amorphous, crystals, solvates,
and hydrates forms, that are fairly tractable from a processing notion. More to the point,
hydrates or solvates are considered as pseudo-polymorphs that may either be an entirely
different crystalline entity or simply incorporated in the parent crystal lattice; however, it is
often possible to remove water or solvate by recrystallization.
Fig. 3. Pharmaceutical solids in different forms.
While the molecular organization of drugs may differ having impact on the development
potential of a drug molecule, the underlying concern for the drug performance is the same.
Herein we tried to meet examples and detailed information on how recrystallization was
used to improve the relative characteristics of drugs.
Park et al. designed a supercritical anti-solvent process for recrystallization of fluconazole, a
triazole antifungal drug which has different polymorphic forms, to modify solid state
characteristics of fluconazole by a range of operating conditions such as temperature,
pressure, and type of solvent. Fluconazole particles were detected by means of differential
scanning calorimetry, thermogravimetry analysis, powder X-ray diffraction, fourier
transform infrared spectroscopy and scanning electron microscopy. Regarding the findings
of the study, polymorphic forms of fluconazole were remarkably affected by the
experimental conditions as the anhydrate form I of drug was obtained at low temperature
and the anhydrate form II was obtained at higher temperatures. On the whole they
suggested that the solid state characteristics of fluconazole, including the polymorphic form
Recrystallization of Drugs: Significance on Pharmaceutical Processing
435
could effectively be managed by altering the operating conditions of the recrystallization
process such as temperature, pressure, and solvent (Park et al., 2007). In another study,
recrystallization of two anti-cancer drugs, erlotinib hydrochloride and fulvestrant was
investigated employing the same method of recrystallization by Tien et al. Polymorph
conversion from the original form B to form E or a prior form A for erlotinib was
demonstrated under appropriate operational conditions which improved the dissolution
rate of the drug. The micronized fulvestrant drug particles showed consistent polymorph
like the original drug, but with differences in crystal habits. They also confirmed the positive
effect of recrystallization in drug modifications (Tien et al., 2010).
Chen et al. evaluated the effects for the type of solvent, temperature and pressure as well as
the solution flow rate on sulfathiazole particle formation using the same supercritical
antisolvent technology for recrystallization. In the optimum operating condition of acetone
as the solvent, temperature at 308 K with 12 MPa pressure and flow rate of 2mL/min the
micronized sulfathiazole including polymorphic forms were obtained. Moreover, it was
determined that various solvents resulted in different polymorphisms where the polymorph
form III changed to form IV when ethanol was employed as the solvent. As a result,
recrystallization of sulfathiazole produced in optimally micronized particles which
exhibited a much narrower particle size distribution with an enhanced in vitro dissolution
rate by 3.2 times to the original form of the drug (Chen et al., 2010).
4. Micro-crystal formation
Apart from the final quality of the drug crystal product in terms of purity, polymorphism,
habit and morphology as well as crystal mechanical strength characterized in
recrystallization procedure, size distributions of the drug particle is another crucial aspect of
the pharmaceutical dosage forms. With reference to the NoyesWhitney equation,
application of a drug substance in a reduced particle size encourages bioavailability of the
poorly water-soluble drug substances owing to the enhanced dissolution rate of micron- or
nano-size drug particles (Chaumeil, 1998). An alternative area where small sized drug
particles are indispensable is the pulmonary drug administration, inhaler drugs, in which
drug powders should have a narrow particle size distribution and a mean particle size of 5
m with almost no particles larger than 10 m. Pharmaceutical dosage forms of these kinds
are supposed to follow low particle agglomeration tendency, sufficient flow properties, and
good batch-to-batch conformity in favor of the relative drug powders (IslamGladki, 2008).
Several techniques could be applied for the preparation of micron sized pharmaceuticals,
such as mechanical comminution of the previously formed larger drug particles by
crushing, grinding, milling and etc. that are the most common ways in this field. However,
these methods not only provide limited opportunity for the control of important drug
characteristics, like size, shape, morphology, surface properties but also ascertain distorted
drug properties in a principally uncontrolled manner (RasenackMuller, 2004). As the
surfaces in mechanically micronized drug powders are not naturally grown as the crystal
cleaves at the crystal face and the surface energy changes, processing properties like
flowability, agglomeration and stickiness to surfaces are the prevalent phenomenon.
Employing milling processes such as jet milling, pearl-ball milling, or high-pressure
homogenization other than influencing the preformulation behavior of drug structures, call
Recrystallization
436
for high energy and manpower being evidence for insufficient coverage of this method for
application in reduced particle size drug production. It seems that these commonly being
used techniques do not meet the ideal way for the production of small sized drug particles.
Since the suitable physicochemical and biopharmaceutical properties of a drug substance
add to the time and cost of drug development, any postern to resolve these problems and
produce small particles of drugs in a controlled process maximizes the opportunity to
succeed in drug product manufacturing. Unlike the former techniques, production of small
particles using controlled production processes such as spray drying, precipitation from
supercritical fluid and recrystallization could be applied for the preparation of properly
characterized micron sized pharmaceuticals. Microcrystallization in which the solubility and
dissolution rate is improved by forming high specific surface area is used for preparation of
drug microcrystals by recrystallization methods to reduce the size of the poorly water-
soluble drug particles.
The usual technologies for recrystallization are fulfilled in this framework and it is to use
solvent change or precipitation method by immediate mixing two liquids in presence of
stabilizing agents. Regardless of the absolute efficiency of the recrystallization approach in
production of large drug crystals, producing the small drug particles is still a sort of a
challenge due to the high surface area of these particles, exerting tendency of a particle
growth. So stabilizing agents would be foremost part in this system preventing particle
growth by stabilizing the high specific surface area of small particles (Lechuga-
BallesterosRodriguez-Hornedo, 1993). Nevertheless, recrystallization still continues to be
one of the important parts of small sized drug production in drug development strategies.
Therefore, microcrystals precipitate in presence of stabilizing protective polymers and a
large and hydrophilized surface would be formed in a one process step having advantages
over traditional milling techniques.
Hence, developing micro-crystallization as an efficient approach that modifies the
biopharmaceutical and technological behavior of drug through selection of the process
variables to reach an optimal pharmaceutical product has evolved to meet drug
development challenges. Exploration on the growing number of publications in domain of
micronized drug particle developing techniques apparently to enhance drug dissolution
rate, considering the widely increasing number of poorly water-soluble drugs affirms the
declaration.
Rasenack et al. prepared microcrystals of a poorly water-soluble drug ECU-01, an anti-
inflammatory drug in preclinical state of development by a precipitation practice in the
presence of stabilizing agents such as gelatin, chitosan, and different types of cellulose ethers
and then spray-drying of the formed dispersion. Considering the low specific surface area of
the nearly cuboid-like form of ECU-01, the aim of the survey was to enhance the drug
dissolution rate by using microcrystals. Precipitation came off through dissolving the drug in
acetone followed by an instant pouring an aqueous solution of the stabilizer into the drug
solution. Via the use of cellulose type ethers as a stabilizer employed in this technique the
thermodynamically unstable small particles were stabilized forming a protective layer on the
crystal surface of the homogeneous microcrystals dispersions. Due to the polymorphic nature
of the drug, the newly formed crystals appeared in a needle-shaped habit, highly increasing
the specific surface area. Consequently, the dissolution rate rose up to 93% after 20 min
Recrystallization of Drugs: Significance on Pharmaceutical Processing
437
compared to the 4% in common drug, indicating large surface of the microcrystals.
Recrystallization of the poorly water-soluble drug ECU-01 was considered as a superior
method which is easy to handle and only entails ordinary equipment (Rasenack et al., 2003).
A nonsteroidal anti-inflammatory drug, indomethacin, which has a hydrophobic and pH-
dependent solubility nature, was developed and studied by Kim et al. through a
microcrystallization technique to improve its physicochemical properties. Microcrystals of
indomethacin was produced using a pH-shift procedure in which the drug was dissolved in
an alkaline water to prepare saturated indomethacin solution thereafter the pH of the
solution was decreased by adding 0.5N hydrochloric acid and stored at 20C for 24 h to
form microcrystal. The findings of the study exhibited similar physicochemical properties
for the microcrystals produced and the standard crystalline powder in X-ray diffraction,
differential scanning calorimetry, and Fourier transform infrared spectroscopy analyses,
exclusive of a lower peak height in X-ray diffraction and somewhat lower melting
temperature. The plate-like with uniform sized microcrystals of indomethacin dissolved
about twice over the standard crystalline powder in the initial phase of dissolution study.
Furthermore, the in vitro biological activity of the indomethacin microcrystals was assessed
in their capacity to inhibit the proliferation of colon cancer cells that showed 20% greater
activity than that of the standard crystalline powder. This view might have implications for
improving the efficiency of chemotherapy in treating patients with malignant neoplasms
using this technique for production of indomethacine microcrystals (Kim et al., 2003).
More recently, Talari et al. evaluated gliclazide microcrystals; a widely used drug for the
treatment of non-insulin-dependent diabetes mellitus which shows a low solubility of 55
mg/L in water and gastric fluids leading to a low dissolution rate and variable
bioavailability. The gliclazide microcrystals were prepared by in situ micronization
techniques based on solvent and pH-shift and were examined for the drug absorption and
pharmacokinetics of GL after oral administration in rats. Compared to the original drug,
scanning electron microscopy showed significant changes in the shape and size of the
prepared crystals using both methods. Recrystallized samples not only showed enhanced
dissolution rates than untreated drug particles but also a reduced particle size of about 30
and 61 times by solvent-change and pH-shift methods were detected for drug crystals,
respectively. Regarding results of the in vivo biological assays for hypoglycemic activity,
microcrystallization of gliclazide using both methods resulted in an increased
pharmacodynamic effect of glucose-lowering in diabetic rats which could be relevant to the
improved dissolution rate of the drug (Talari et al., 2010).
Concisely, the microcrystallization of the drug particulates which has an effect on crystal
habit, could also improve the drug absorption characteristics and the subsequent drug
bioavailability.
5. Lyotropic liquid crystals formation
Liquid crystals (LCs) are a state of matter that has properties between those of a
conventional liquid and those of a solid crystal. For instance, an LC may flow like a liquid,
but its molecules may be oriented in a crystal-like way. There are many different types of LC
phases, which can be distinguished by their different optical properties. When viewed
under a microscope using a polarized light source, different liquid crystal phases will
Recrystallization
438
appear to have distinct textures. The contrasting areas in the textures correspond to domains
where the LC molecules are oriented in different directions. Within a domain, however, the
molecules are well ordered. LC materials may not always be in an LC phase (just as water
may turn into ice or steam).
Liquid crystals can be divided into thermotropic, lyotropic and metallotropic phases.
Thermotropic and lyotropic LCs consist of organic molecules. Thermotropic LCs exhibit a
phase transition into the LC phase as temperature is changed. Lyotropic LCs exhibit phase
transitions as a function of both temperature and concentration of the LC molecules in a
solven . Metallotropic LCs are composed of both organic and inorganic molecules; their LC
transition depends not only on temperature and concentration, but also on the inorganic-
organic composition ratio.
Examples of liquid crystals can be found both in the natural world and in technological
applications. Most modern electronic displays are liquid crystal based. Lyotropic liquid-
crystalline phases are abundant in living systems. For example, many proteins and cell
membranes are LCs. Other well-known LC examples are solutions of soap and various
related detergents, as well as the tobacco mosaic virus.
A lyotropic liquid crystal consists of two or more components that exhibit liquid-crystalline
properties in certain concentration ranges. In the lyotropic phases, solvent molecules fill the
space around the compounds to provide fluidity to the system. In contrast to thermotropic
liquid crystals, these lyotropics have another degree of freedom of concentration that
enables them to induce a variety of different phases.
A compound, which has two immiscible hydrophilic and hydrophobic parts within the
same molecule, is called an amphiphilic molecule. Many amphiphilic molecules show
lyotropic liquid-crystalline phase sequences depending on the volume balances between the
hydrophilic part and hydrophobic part. These structures are formed through the micro-
phase segregation of two incompatible components on a nanometer scale. Soap is an
everyday example of a lyotropic liquid crystal.
Fig. 4. Structure of lyotropic liquid crystal. The red heads of surfactant molecules are in
contact with water, whereas the tails are immersed in oil (blue): bilayer (left) and micelle
(right)
Recrystallization of Drugs: Significance on Pharmaceutical Processing
439
The content of water or other solvent molecules changes the self-assembled structures. At very
low amphiphile concentration, the molecules will be dispersed randomly without any
ordering. At slightly higher (but still low) concentration, amphiphilic molecules will
spontaneously assemble into micelles or vesicles. This is done so as to 'hide' the hydrophobic
tail of the amphiphile inside the micelle core, exposing a hydrophilic (water-soluble) surface to
aqueous solution. These spherical objects do not order themselves in solution, however. At
higher concentration, the assemblies will become ordered. A typical phase is a hexagonal
columnar phase, where the amphiphiles form long cylinders (again with a hydrophilic surface)
that arrange themselves into a roughly hexagonal lattice. This is called the middle soap phase.
At still higher concentration, a lamellar phase (neat soap phase) may form, wherein extended
sheets of amphiphiles are separated by thin layers of water. For some systems, a cubic (also
called viscous isotropic) phase may exist between the hexagonal and lamellar phases, wherein
spheres are formed that create a dense cubic lattice. These spheres may also be connected to
one another, forming a bicontinuous cubic phase (VroegeLekkerkerker, 1992).
The objects created by amphiphiles are usually spherical (as in the case of micelles), but may
also be disc-like (bicelles), rod-like, or biaxial (all three micelle axes are distinct). These
anisotropic self-assembled nano-structures can then order themselves in much the same
way as thermotropic liquid crystals do, forming large-scale versions of all the thermotropic
phases (such as a nematic phase of rod-shaped micelles).
For some systems, at high concentrations, inverse phases are observed. That is, one may
generate an inverse hexagonal columnar phase (columns of water encapsulated by
amphiphiles) or an inverse micellar phase (a bulk liquid crystal sample with spherical water
cavities).
A generic progression of phases, going from low to high amphiphile concentration, is:
Discontinuous cubic phase (micellar cubic phase), Hexagonal phase (hexagonal columnar
phase) (middle phase), Lamellar phase, Bicontinuous cubic phase, Reverse hexagonal
columnar phase, and Inverse cubic phase (Inverse micellar phase)
Even within the same phases, their self-assembled structures are tunable by the
concentration: for example, in lamellar phases, the layer distances increase with the solvent
volume. Since lyotropic liquid crystals rely on a subtle balance of intermolecular
interactions, it is more difficult to analyze their structures and properties than those of
thermotropic liquid crystals. Similar phases and characteristics can be observed in
immiscible diblock copolymers.
Lyotropic liquid crystals transitions occur with influence of solvents or recrystallization
method. Lyotropic liquid crystals occur as a result of solvent induced aggregation of the
constituent mesogens into micellar structure. This types of liquid crystalline states were
mostly used for designing the sustain release drug delivery system. It is also used for
improving the solubility and stability of insoluble drugs by incorporating it into micellar
structure of liquid crystals (Lechuga-Ballesteros et al., 2003).
6. Spherical crystallization
Among the particles designed for solid pharmaceutical dosage forms, tablet supports for
the half of all oral drug delivery system and 70% of the all pharmaceutical preparations
Recrystallization
440
produced. Direct tabletting, simple mixing and compressing drug powders have been
widely applied to a large number of drugs on the industrial scale as an excellent
technique. A successful tabletting and also compression of any drug material is reliant on
the micromeritic properties of the drug crystals. In this context, crystals of needle-shaped
or plated-shaped are challengeable to be handled with, due to the poor flowability of
these crystals (Kaerger et al., 2004). In 1984 Kawashima et al. introduced crystal
agglomeration with controlled properties as spherical crystallization to the
pharmaceutical manufacturing and expressed that the spherically dense agglomerates
were suitable for direct tabletting. Accordingly, crystallization and agglomeration of the
drug substance particles concurrently in one step to transform crystals directly into
compacted spherical form during the recrystallization process is defined as spherical
crystallization (Kawashima, 1984). Spherical crystals could be established through two
different techniques, either by typical spherical crystallization technique or non typical
spherical crystallization technique (Nokhodchi et al., 2007). Non typical spherical
crystallization technique might be regarded as the traditional crystallization process like
salting-out, cooling, precipitation, whereas, the typical spherical crystallization is a three-
solvent system employing three solvents; one is the drug dissolution medium known as
the good solvent, another is a medium which partially dissolves the drug and has wetting
feature that is named bridging liquid, and the last one is immiscible with the drug
substance recognized as the bad solvent (Nokhodchi, Maghsoodi et al., 2007; Mahanty et
al., 2010).
On average, spherical crystallization techniques are believed to be promising techniques
in which the drug crystals are modified using different solvents for direct compressible
spherical agglomerates, which can save money and time for tabletting. These approaches
not only helps to achieve good flowability, compressibility and micromeritic properties of
the drug substances, but also it is known to improve the the wettability, bioavailability,
and dissolution rate of some poorly soluble drugs (Kawashima, Handa et al., 1986; Di
Martino et al., 1999; NokhodchiMaghsoodi, 2008). Besides, drug materials produced by
the spherical crystallization technique result in the economical process in the
development of the solid dosage forms for scaling up to a commercial level since it
provides reduced time and cost by enabling faster operation, less machinery and fewer
personnel.
Seeing that, magnesium aspartate and acetylsalicylic acid crystals in common are
tetragonal and prism-shaped with different sizes they show poor flowability and
compactibility properties indicating the crystal habit and the electrostatic charge. Szabo-
Revesz et al. developed magnesium aspartate and acetylsalicylic acid via spherical
crystallization since they are used in direct tablet-making and capsule-filling, the particle
size and the spherical form are fundamental in view of their processibility. They
prepared drug agglomerates through non-typical (magnesium aspartate) and typical
(acetylsalicylic acid) spherical crystallization techniques. Crystal agglomerates of these
drugs produced by these different spherical crystallization techniques created
opportunity for a comparison between the results obtained. Considering the findings of
the study, a higher initial cooling rate and a lower stirring rate were favorable in
producing crystal agglomerates. The growth of particle size and the spherical form as
well as the associated decreased specific surface of the magnesium aspartate and
Recrystallization of Drugs: Significance on Pharmaceutical Processing
441
acetylsalicylic acid crystal agglomerates created better compactibility and cohesivity
characteristics than the control samples in a softer flow time, and a higher bulk density.
According to the authors both types of spherical crystallization (non-typical and typical)
can be effectively used not only for spherical particle forming but also for size growing
of drug materials that are pressed directly into tablets or made into filled capsules
without excipients (Szabo-Revesz et al., 2002).
Celecoxib, a non-steroidal anti-inflammatory drug which is the first selective
cyclooxygenase-2 inhibitor used in the treatment of osteoarthritis and rheumatoid
arthritis exhibits poor flow and compression characteristics as well as incomplete and
poor oral bioavailability due to its low aqueous solubility. Variety of papers in this case is
representative of the suitability of this drug for spherical crystallization process to
enhance the flow, compressibility and solubility properties of seems to be a beneficial
objective to improve the following therapeutic efficacy of celecoxib (Banga et al., 2007).
Paradkar et al. improved the micromeritic and compressional properties of celecoxib by a
spherical crystallization process using the solvent change method. Selection of the
solvents depended on the miscibility of solvents and also the solubility of drug in the
solvents, which candidated acetone as the good solvent, dichloromethane as the bridging
liquid, and water as the bad solvent. A solution of celecoxib in acetone was added to a
solution of hydroxy propyl methylcellulose in dichloromethane. Drug was crystallized by
adding the solution to wall-baffled vessel containing distilled water followed by a
continuous mixing in a controlled speed style creating agglomerated spherical crystals.
Celecoxib agglomerates exhibited satisfactory micromeritic, mechanical, and
compressional properties demonstrating comparable in vitro drug release performance
with the marketed capsule formulation (Paradkar et al., 2002). Elsewhere, Gupta et al.
prepared spherical crystals of celecoxib using a more hydrophilic polymer,
polyvinylpyrrolidone K-30 (PVP) and acetone, water and chloroform as solvent, non-
solvent and bridging liquid, respectively. The agglomerates were determined by
differential scanning calorimetry, X-ray diffraction, IR spectroscopic and scanning
electron microscopy. They showed that the crystals possessed a good spherical shape with
smooth and regular surface exhibiting significantly improved micromeritic properties
compared to pure the drug. Moreover, the aqueous solubility and dissolution rate of the
drug from crystals was notably increased nearly two times, with an increase in PVP
concentration. In general, this technique may be applicable for producing oral solid
dosage forms of other drugs with improved dissolution rate and oral bioavailability
(Gupta et al., 2007). Correspondingly, use of these systems has the potential to facilitate
drug development by saving valuable time in selecting the optimal physical or chemical
characteristics of a given compound.
On the whole, lessening the risk of drug process modifications and providing the
opportunity to gain more comprehensive rational property coverage would be established if
such information is established through recrystallization processes at an early stage of drug
developments. Eventually, it is of the utmost importance to strictly monitor the processing
of drug substances with regards to the different crystal habits of drug materials, as well as to
obtain a comprehensive understanding of the physical and chemical stability of these
polymorphic states.
Recrystallization
442
7. References
Adhiyaman, R. & Basu, S.K. (2006). Crystal modification of dipyridamole using different
solvents and crystallization conditions. International Journal of Pharmaceutics,
Vol.321, No.1-2, pp. 27-34, ISSN 03785173
Adibkia, K., Hamedeyazdan, S. & Javadzadeh, Y. (2011). Drug release kinetics and
physicochemical characteristics of floating drug delivery systems. Expert Opin Drug
Deliv, Vol.8, No.7, pp. 891-903, ISSN 1744-7593
Banga, S., Chawla, G., Varandani, D., Mehta, B.R. & Bansal, A.K. (2007). Modification of the
crystal habit of celecoxib for improved processability. J Pharm Pharmacol, Vol.59,
No.1, pp. 29-39, ISSN 0022-3573
Blagden, N., de Matas, M., Gavan, P.T. & York, P. (2007). Crystal engineering of active
pharmaceutical ingredients to improve solubility and dissolution rates. Adv Drug
Deliv Rev, Vol.59, No.7, pp. 617-630, ISSN 0169-409X
Carino, S.R., Sperry, D.C. & Hawley, M. (2006). Relative bioavailability estimation of
carbamazepine crystal forms using an artificial stomach-duodenum model. J Pharm
Sci, Vol.95, No.1, pp. 116-125, ISSN 0022-3549
Chaumeil, J.C. (1998). Micronization: a method of improving the bioavailability of poorly
soluble drugs. Methods Find Exp Clin Pharmacol, Vol.20, No.3, pp. 211-215, ISSN
0379-0355
Chen, J. & Trout, B.L. (2008). Computational study of solvent effects on the molecular
self-assembly of tetrolic acid in solution and implications for the polymorph
formed from crystallization. J Phys Chem B, Vol.112, No.26, pp. 7794-7802, ISSN
1520-6106
Chen, Y.-M., Tang, M. & Chen, Y.-P. (2010). Recrystallization and micronization of
sulfathiazole by applying the supercritical antisolvent technology. Chemical
Engineering Journal, Vol.165, No.1, pp. 358-364, ISSN 13858947
Chow, A.H.L., Hsia, C.K., Gordon, J.D., Young, J.W.M. & Vargha-Butleff, E.I. (1995).
Assessment of wettability and its relationship to the intrinsic dissolution rate of
doped phenytoin crystals. International Journal of Pharmaceutics, Vol.126 pp. 21-28,
ISSN 0378-5173
Dahlberg, C., Dvinskikh, S.V., Schuleit, M. & Furo, I. (2011). Polymer Swelling, Drug
Mobilization and Drug Recrystallization in Hydrating Solid Dispersion Tablets
Studied by Multinuclear NMR Microimaging and Spectroscopy. Mol Pharm, Vol.8,
No.4, pp. 1247-1256, ISSN 1543-8392
Di Martino, P., Barthelemy, C., Piva, F., Joiris, E., Palmieri, G.F. & Martelli, S. (1999).
Improved dissolution behavior of fenbufen by spherical crystallization. Drug Dev
Ind Pharm, Vol.25, No.10, pp. 1073-1081, ISSN 0363-9045
Grzesiak, A.L., Lang, M., Kim, K. & Matzger, A.J. (2003). Comparison of the four anhydrous
polymorphs of carbamazepine and the crystal structure of form I. J Pharm Sci,
Vol.92, No.11, pp. 2260-2271, ISSN 0022-3549
Gupta, V.R., Mutalik, S., Patel, M.M. & Jani, G.K. (2007). Spherical crystals of celecoxib to
improve solubility, dissolution rate and micromeritic properties. Acta Pharm,
Vol.57, No.2, pp. 173-184, ISSN 1330-0075
Recrystallization of Drugs: Significance on Pharmaceutical Processing
443
Harbury, L. (1947). Solubility and melting point as functions of particle size; the induction
period of crystallization. J Phys Colloid Chem, Vol.51, No.2, pp. 382-391, ISSN 0092-
7023
Heng, J.Y., Bismarck, A. & Williams, D.R. (2006). Anisotropic surface chemistry of
crystalline pharmaceutical solids. AAPS PharmSciTech, Vol.7, No.4, pp. 84, ISSN
1530-9932
Islam, N. & Gladki, E. (2008). Dry powder inhalers (DPIs)--a review of device reliability and
innovation. Int J Pharm, Vol.360, No.1-2, pp. 1-11, ISSN 0378-5173
Jamali, F. & Mitchell, A.G. (1973). The recrystallization and dissolution of acetylsalicylic
acid. Acta Pharm Suec, Vol.10, No.4, pp. 343-352, ISSN 0001-6675
Javadzadeh, Y., Mohammadi, A., Khoei, N.S. & Nokhodchi, A. (2009). Improvement of
physicomechanical properties of carbamazepine by recrystallization at different pH
values. Acta Pharm, Vol.59, No.2, pp. 187-197, ISSN 1330-0075
Jozwiakowski, M.J., Nguyen, N.A., Sisco, J.M. & Spancake, C.W. (1996). Solubility behavior
of lamivudine crystal forms in recrystallization solvents. J Pharm Sci, Vol.85, No.2,
pp. 193-199, ISSN 0022-3549
Kaerger, J.S., Edge, S. & Price, R. (2004). Influence of particle size and shape on flowability
and compactibility of binary mixtures of paracetamol and microcrystalline
cellulose. Eur J Pharm Sci, Vol.22, No.2-3, pp. 173-179, ISSN 0928-0987
Kawashima, Y. (1984). Development of spherical crystallization technique and its
application to pharmaceutical systems. Arch Pharm. Res., Vol.7, No.2, pp. 145-
151
Kawashima, Y. (2003). Improved flowability and compactibility of spherically
agglomerated crystals of ascorbic acid for direct tableting designed by spherical
crystallization process. Powder Technology, Vol.130, No.1-3, pp. 283-289, ISSN
00325910
Kawashima, Y., Handa, T., Takeuchi, H., Okumura, M., Katou, H. & Nagata, O. (1986).
Crystal modification of phenytoin with polyethylene glycol for improving
mechanical strength, dissolution rate and bioavailability by a spherical
crystallization technique. Chem Pharm Bull (Tokyo), Vol.34, No.8, pp. 3376-3383,
ISSN 0009-2363
Kim, S.T., Kwon, J.H., Lee, J.J. & Kim, C.W. (2003). Microcrystallization of indomethacin
using a pH-shift method. Int J Pharm, Vol.263, No.1-2, pp. 141-150, ISSN 0378-
5173
Kobayashi, Y., Ito, S., Itai, S. & Yamamoto, K. (2000). Physicochemical properties and
bioavailability of carbamazepine polymorphs and dihydrate. Int J Pharm, Vol.193,
No.2, pp. 137-146, ISSN 0378-5173
Krishnaiah S.R., Y. (2010). Pharmaceutical Technologies for Enhancing Oral Bioavailability
of Poorly Soluble Drugs. Journal of Bioequivalence & Bioavailability, Vol.02, No.02, pp.
28-36, ISSN 0975-0851
Krishnaiah, Y.S.R. (2010). Pharmaceutical technologies for enhancing oral bioavailability of
poorly soluble drugs. Journal of Bioequivalence & Bioavailability, Vol.2, No.2, pp. 028-
036, ISSN 0975-0851
Lechuga-Ballesteros, D., Abdul-Fattah, A., Stevenson, C.L. & Bennett, D.B. (2003). Properties
and stability of a liquid crystal form of cyclosporine-the first reported naturally
Recrystallization
444
occurring peptide that exists as a thermotropic liquid crystal. J Pharm Sci, Vol.92,
No.9, pp. 1821-1831, ISSN 0022-3549
Lechuga-Ballesteros, D. & Rodriguez-Hornedo, N. (1993). Growth and morphology of L-
alanine crystals: influence of additive adsorption. Pharm Res, Vol.10, No.7, pp. 1008-
1014, ISSN 0724-8741
Liebenberg, W., de Villiers, M.M., Wurster, D.E., Swanepoel, E., Dekker, T.G. & Lotter, A.P.
(1999). The effect of polymorphism on powder compaction and dissolution
properties of chemically equivalent oxytetracycline hydrochloride powders. Drug
Dev Ind Pharm, Vol.25, No.9, pp. 1027-1033, ISSN 0363-9045
Llinas, A., Burley, J.C., Box, K.J., Glen, R.C. & Goodman, J.M. (2007). Diclofenac solubility:
independent determination of the intrinsic solubility of three crystal forms. J Med
Chem, Vol.50, No.5, pp. 979-983, ISSN 0022-2623
Lobenberg, R. & Amidon, G.L. (2000). Modern bioavailability, bioequivalence and
biopharmaceutics classification system. New scientific approaches to
international regulatory standards. Eur J Pharm Biopharm, Vol.50, No.1, pp. 3-12,
ISSN 0939-6411
Maghsoodi, M., Hassan-Zadeh, D., Barzegar-Jalali, M., Nokhodchi, A. & Martin, G. (2007).
Improved compaction and packing properties of naproxen agglomerated crystals
obtained by spherical crystallization technique. Drug Dev Ind Pharm, Vol.33, No.11,
pp. 1216-1224, ISSN 0363-9045
Mahanty, S., Sruti, J., Niranjan Patra, C. & Bhanoji Rao, M.E. (2010). Particle design of
drugs by spherical crystallization techniques. International Journal of
Pharmaceutical Sciences and Nanotechnology, Vol.3, No.2, pp. 912-918, ISSN 0974
9446
Mirmehrabi, M. & Rohani, S. (2005). An approach to solvent screening for crystallization of
polymorphic pharmaceuticals and fine chemicals. J Pharm Sci, Vol.94, No.7, pp.
1560-1576, ISSN 0022-3549
Nokhodchi, A., Bolourtchian, N. & Dinarvand, R. (2003). Crystal modification of phenytoin
using different solvents and crystallization conditions. Int J Pharm, Vol.250, No.1,
pp. 85-97, ISSN 0378-5173
Nokhodchi, A. & Maghsoodi, M. (2008). Preparation of spherical crystal agglomerates of
naproxen containing disintegrant for direct tablet making by spherical
crystallization technique. AAPS PharmSciTech, Vol.9, No.1, pp. 54-59, ISSN 1530-
9932
Nokhodchi, A., Maghsoodi, M. & Hassanzadeh, D. (2007). An Improvement of
Physicomechanical Properties of Carbamazepine Crystals. Iranian Journal of
Pharmaceutical Research, Vol.6, No.2, pp. 83-89, ISSN 1735-0328
Paradkar, A.R., Pawar, A.P., Chordiya, J.K., Patil, V.B. & Ketkar, A.R. (2002). Spherical
crystallization of celecoxib. Drug Dev Ind Pharm, Vol.28, No.10, pp. 1213-1220, ISSN
0363-9045
Park, H.J., Kim, M.S., Lee, S., Kim, J.S., Woo, J.S., Park, J.S. & Hwang, S.J. (2007).
Recrystallization of fluconazole using the supercritical antisolvent (SAS) process.
Int J Pharm, Vol.328, No.2, pp. 152-160, ISSN 0378-5173
Perlovich, G.L., Strakhova, N.N., Kazachenko, V.P., Volkova, T.V., Tkachev, V.V.,
Schaper, K.J. & Raevsky, O.A. (2008). Sulfonamides as a subject to study
Recrystallization of Drugs: Significance on Pharmaceutical Processing
445
molecular interactions in crystals and solutions: sublimation, solubility,
solvation, distribution and crystal structure. Int J Pharm, Vol.349, No.1-2, pp. 300-
313, ISSN 0378-5173
Rasenack, N., Hartenhauer, H. & Muller, B.W. (2003). Microcrystals for dissolution rate
enhancement of poorly water-soluble drugs. Int J Pharm, Vol.254, No.2, pp. 137-145,
ISSN 0378-5173
Rasenack, N. & Muller, B.W. (2004). Micron-size drug particles: common and novel
micronization techniques. Pharm Dev Technol, Vol.9, No.1, pp. 1-13, ISSN 1083-
7450
Rodriguez-Spong, B., Price, C.P., Jayasankar, A., Matzger, A.J. & Rodriguez-Hornedo,
N. (2004). General principles of pharmaceutical solid polymorphism: a
supramolecular perspective. Adv Drug Deliv Rev, Vol.56, No.3, pp. 241-274,
ISSN 0169-409X
Seton, L., Roberts, M. & Ur-Rehman, F. (2010). Compaction of recrystallised ibuprofen.
Chemical Engineering Journal, Vol.164, No.2-3, pp. 449-452, ISSN 1385-8947
Sinclair, W., Leane, M., Clarke, G., Dennis, A., Tobyn, M. & Timmins, P. (2011). Physical
stability and recrystallization kinetics of amorphous ibipinabant drug product by
fourier transform raman spectroscopy. J Pharm Sci, pp., ISSN 1520-6017
Suzuki, T., Araki, T., Kitaoka, H. & Terada, K. (2010). Studies on mechanism of thermal
crystal transformation of sitafloxacin hydrates through melting and
recrystallization, yielding different anhydrates depending on initial crystalline
forms. Int J Pharm, Vol.402, No.1-2, pp. 110-116, ISSN 1873-3476
Szabo-Revesz, P., Hasznos-Nezdei, M., Farkas, B., Goczo, H., Pintye-Hodi, K. & Eros, I.
(2002). Crystal growth of drug materials by spherical crystallization. Journal of
Crystal Growth, Vol.237, No.239, pp. 22402245, ISSN 0022-0248
Talari, R., Varshosaz, J., Mostafavi, S.A. & Nokhodchi, A. (2010). Gliclazide microcrystals
prepared by two methods of in situ micronization: pharmacokinetic studies in
diabetic and normal rats. AAPS PharmSciTech, Vol.11, No.2, pp. 786-792, ISSN 1530-
9932
Tian, F., Sandler, N., Aaltonen, J., Lang, C., Saville, D.J., Gordon, K.C., Strachan, C.J.,
Rantanen, J. & Rades, T. (2007). Influence of polymorphic form, morphology, and
excipient interactions on the dissolution of carbamazepine compacts. J Pharm Sci,
Vol.96, No.3, pp. 584-594, ISSN 0022-3549
Tien, Y.-C., Su, C.-S., Lien, L.-H. & Chen, Y.-P. (2010). Recrystallization of erlotinib
hydrochloride and fulvestrant using supercritical antisolvent process. The Journal of
Supercritical Fluids, Vol.55, No.1, pp. 292-299, ISSN 0896-8446
Tiwary, A.K. (2001). Modification of crystal habit and its role in dosage form performance.
Drug Dev Ind Pharm, Vol.27, No.7, pp. 699-709, ISSN 0363-9045
Vroege, G.J. & Lekkerkerker, H.N.W. (1992). Phase transitions in lyotropic colloidal
and polymer liquid crystals. Rep. Fmg. Phys., Vol.55, pp. 1241-1309, ISSN 0034-
4885
Wagner, D., Glube, N., Berntsen, N., Tremel, W. & Langguth, P. (2003). Different
dissolution media lead to different crystal structures of talinolol with impact on
its dissolution and solubility. Drug Dev Ind Pharm, Vol.29, No.8, pp. 891-902, ISSN
0363-9045
Recrystallization
446
Whler, F. & Liebig, J. (1832). Untersuchungen ber das Radikal der Benzoesure. Annalen
der Pharmacie, Vol.3, No.3, pp. 249-282, ISSN 0075-4617
Yeo, S., Kim, M. & Lee, J. (2003). Recrystallization of sulfathiazole and chlorpropamide using
the supercritical fluid antisolvent process. The Journal of Supercritical Fluids, Vol.25,
No.2, pp. 143-154, ISSN 0896-8446
19
Crystal Forms of Anti-HIV Drugs:
Role of Recrystallization
Renu Chadha, Poonam Arora, Anupam Saini and Swati Bhandari
University Institute of Pharmaceutical Sciences, Panjab University, Chandigarh,
India
1. Introduction
Understanding and controlling the solid-state chemistry of active pharmaceutical
ingredients (APIs) is an important aspect of drug development process. APIs can exist in a
variety of distinct solid forms, including polymorphs, solvates, hydrates, salts, cocrystals
and amorphous solids. Most APIs are purified and isolated by crystallization from an
appropriate solvent during the final step in the synthetic process. A large number of factors
can influence the crystal nucleation and growth during this process, including the
composition of the crystallization medium and the processes used to generate
supersaturation and promote crystallization. For development of a pharmaceutical product,
it is generally accepted that the stable form should be identified and chosen for
development. However, the stable crystal form of the parent compound may exhibit
inadequate solubility or dissolution rate resulting in poor oral absorption, particularly for
water insoluble compounds whereas a metastable form might have advantageous
properties. Moreover, the metastable polymorphs constitute local minima in the energy
landscape (Figure 1) with the thermodynamically stable form being the absolute minimum
at a given temperature and pressure. Thus, the search for absolute minimum and energy
differences between the local minima of the drug substances is the goal of material and
formulation scientists in the pharmaceutical industry. While significant efforts are made by
drug development groups to identify and characterize thermodynamically stable crystal
forms early in development, there are many instances where new crystal forms have been
discovered later in development process. The late emergence of thermodynamically stable
crystal form is often explained by Ostwalds law of stages which states that the least stable
crystal form is likely to crystallize first. Metastable forms appear first during crystallization
process as their crystallization kinetics is faster than those of stable forms but eventually
transform into a stable form. It is important to study the transformations, because the
sudden appearance or disappearance of a polymorphic form in pharmaceutical products can
lead to serious consequences.
Therefore, it is of utmost importance to control the crystal formation and produce a desired
form. Discovery and characterization of the diversity of solid forms of a drug substance
provide options from which to select a form that exhibits the appropriate balance of the
critical properties for development into the drug product. Lately, the crystal engineering
approaches utilizing high throughput techniques have been applied to crystalline materials
Recrystallization
448
to fruitfully generate various crystal forms of pharmaceutical compounds. The ability to
engineer pharmaceutical materials with suitable solubility characteristics whilst maintaining
suitable physical and chemical stability provides a driving force to utilize modern analytical
tools to generate new crystal modifications.
Drugs with multicomponent crystalline phases such as cocrystals also carry desired drug
properties similar to single-component polymorphs. The pharmaceutical cocrystals is
defined as multicomponent molecular complex where one of the components is an
investigational or marketed drug molecule (the active pharmaceutical ingredient or API)
and the second component (the coformer) is a safe chemical for human consumption
selected from the GRAS list of the US FDA (generally regarded as safe additive chemicals by
the Food and Drugs Administration). The two components are present in a definite
stoichiometric ratio and interact through noncovalent interactions, predominantly hydrogen
bonds. Cocrystals have been found to offer an attractive platform to improve the solubility
and dissolution rate of pharmaceuticals without compromising on the stability of the solid
form. Many pharmaceutical companies are working actively on cocrystals and this is
reflected by the growing number of publications and patent applications for co-crystals in
recent years.
Fig. 1. Energy landscape for polymorphic forms.
The identification and characterization of diverse crystalline forms of the drug substances
has become imperative for the solid state chemists in order to select the appropriate form
to ensure that the product performance with respect to manufacturability, stability and
bioavailability remains unchanged. Therefore, this chapter deals with the effect of
crystalline state of pharmaceutical material on its physicochemical properties
demonstrated through several case studies describing the phenomenon of polymorphism
and a particular attention will also be paid to cocrystallization which is emerging as an
important technique to generate crystal forms with improved physicochemical properties
(Scheme 1).
Crystal Forms of Anti-HIV Drugs: Role of Recrystallization
449
Scheme 1. Scheme representing the path followed by drugs from nucleation to production
1.1 Effect of crystal forms on physicochemical properties
It is well established that different solid forms of drug molecules exist and can affect
pharmaceutical drug products with respect to physicochemical properties. It is the variation
in the physical and chemical properties of drug molecules that makes polymorphism such a
potentially important issue for the pharmaceutical industry. The APIs with abundant
hydrogen bond sites and molecular flexibility may be manipulated by proper choice of
solvent to form a specific crystal form that has different arrangements or conformations of
the molecule in the crystal lattice. Changing the arrangement of the molecules in the crystal
lattice changes the solid state properties affecting its solubility, stability, dissolution rate and
bioavailability. Different crystal forms can have different rates of uptake in the body,
leading to lower or higher biological activity than desired. As a result, significant effort is
placed in identifying suitable solid forms of drug substances for use in pharmaceutical drug
products.
1.2 Crystallization process
The crystallization process of polymorphous crystals is concerned with evolution of
crystalline state from solution or melts and is composed of competitive nucleation and
growth. However, crystallization is a complex process which starts with the appearance of
most soluble form and ends with transformation to stable form. Alternatively, the
nucleation of the stable form can be initiated by the dissolution of this metastable form and
growth of the stable form continues until the solubility of stable forms is reached. In many
instances the metastable form with desirable properties may precipitate out which is stable
for years. To selectively crystallize polymorphs, the mechanism of each elementary step in
crystallization process needs to be clear in relation to operational conditions and the key
controlling factors. The nucleation process is the most important for control of
polymorphous crystallization. Thus control over solid form throughout the drug
development process is of paramount importance.
A number of factors (Figure 2) affect polymorphic behavior of a pharmaceutical solid and
type of solvent has a major factor in polymorphic selectivity and crystal morphology.
1.3 Effect of solvent on crystal form
Polymorph selectivity is primarily based on the polarity of the solvent. Thus a systematic
approach for selecting the right solvent is beneficial for better experimental design in control
Recrystallization
450
of crystallization. This effect arises from the solventsolute interaction at the molecular level
and have been explained by few of the researchers. The solventsolute interactions during
cluster formation for nucleation and growth significantly affect the ultimate crystal structure
and morphology. If the solventsolute are strongly bonded at a special crystal face, the rate-
limiting step of growth would be the removal of the solvent from that face. In this case, the
bonded surface grows slowly or does not grow.
However, Threlfall showed that if crystallization occurs in a region that is supersaturated
with respect to one polymorph (the less soluble form) and under-saturated with respect to
the other one, the solvent has no influence on the nucleation of the polymorphs and the
thermodynamics will lead the process toward the production of the less soluble polymorph
(Threlfall T., 2000). Moreover, the thermodynamically stable polymorph is the most stable
form irrespective of the type of the solvent used. Since the thermodynamic stability of
different polymorphs does not change with type of the solvent, then the solvent effect on
polymorphism is attributed to the kinetic parameters.
Fig. 2. Factors which affect the crystallization of a drug substance
1.4 Other factors
Within the same solvent system, many factors are known to influence the crystal habit
including supersaturation, cooling rate and agitation. At a constant temperature,
supersaturation has a direct effect on the nucleation rate. As the supersaturation is
increased, the rate of nuclei formation is greater than crystal growth and growth occurs
mainly in one direction, producing elongated crystals. On the other hand, when there is
lesser degree of supersaturation, solute solvent interactions are insignificant, producing
platy crystals.
Crystal Forms of Anti-HIV Drugs: Role of Recrystallization
451
Similarly, rate of cooling alters the crystal habit by its influence on degree of supersaturation.
Crystallization at a slower cooling rate produces more symmetric crystals compared with
faster cooling. During faster cooling, nucleation is faster than crystal growth rate; therefore,
many small crystals appear instead of few crystals growing to sufficiently larger size.
Agitation has also an important effect on the process of crystallization. The aspect ratio
(ratio of horizontal maximum and vertical maximum distance of particle) is highest for
unstirred conditions than during stirring. The crystals obtained under stirring conditions are
fine since stirring facilitates the rate of nucleation by an even distribution of the solute
molecules in the solvent. Increased nucleation rate is the result of collision of initial crystals
with the stirrer and formation of smaller seeds for further crystallization. Additionally,
stirring can also break larger crystals to smaller ones. Thus external appearance of a crystal
can be altered by changing the growth environment to suit the requirements.
2. Polymorphism in anti-HIV drugs
This chapter throws light on the different crystalline forms reported so far for nevirapine,
efavirenz, lamivudine, stavudine and zidovudine. Preparation and isolation methods,
structural characterization and properties of polymorphic/solvatomorphic/cocrystal
systems as well as phase transformations are illustrated.
2.1 Ritonavir
A number of studies have successfully demonstrated the appearance of different crystalline
forms of some anti-HIV drugs upon recrystallization. An early example being that of ritonavir,
marketed as Norvir. The late emergence of a thermodynamically more stable form (Form II)
which unexpectedly precipitated from the semisolid capsule formulation led to the removal of
the product from the market. The new crystal form (form II) appeared after conversion of
metastable crystalline form I (Chemburkar et al., 2000; Bauer et al., 2001; Desikan et al., 2005;
Miller et al., 2005). Ritonavir polymorphism was investigated using solid state spectroscopy
and microscopy techniques including solid state NMR, near infrared spectroscopy, powder X-
ray diffraction and single crystal X-ray analysis. Ritonavir was found to exhibit conformational
polymorphism with two unique crystal lattices having significantly different solubility
properties. An unusual conformation was found for form II that results in a strong hydrogen
bonding network. Although the polymorph (form II) corresponding to the cis conformation
has a more stable packing arrangement, however, nucleation, even in the presence of form II
seeds, is energetically unfavored except in highly supersaturated solutions. The coincidence of
a highly supersaturated solution and a probable heterogeneous nucleation resulted in the
sudden appearance of the more stable form II polymorph.
Form Melting point, C Hfus, J/g Solid state structure
I* 122 78.2 Monoclinic
II* 122 87.8 Orthorhombic
III 7882 60.3 Monoclinic
IV 116 59.8 Not assigned
V 97 32.0 Monoclinic
*Bauer et al.
Table 1. Comparison of physical parameters of ritonavir crystal forms
Recrystallization
452
This polymorphic shift in ritonavir illustrated the need for early and comprehensive
identification of solid-form diversity of this API. The polymorphic behavior of ritonavir was
explored by Morissette et al (Morissette et al., 2003) in 2003 using CrystalMax, a high-
throughput crystallization platform, with the aim of finding known and novel crystal forms
of the drug molecule. Three additional crystalline forms of ritonavir were discovered when
about 2,000 screening experiments were carried out (Table 1). These forms were found along
with both known forms I and II, which were obtained from previously unreported solvent
mixtures. Form III is a crystalline formamide solvate that converts to form V, a previously
unknown hydrated phase, upon exposure to aqueous medium. Form V which is a trihydrate
obtained from exposing the form III to aqueous conditions, in turn converts spontaneously
to needle-like form I crystals. The process of preparing form I from III is an unusual route to
a disappearing polymorph and provides a novel strategy for control of particle size and
morphology. Form IV is a true, unsolvated, metastable previously unreported polymorph of
ritonavir. Optical imaging (Figure 3) and in situ Raman spectroscopy were used to
characterize newly formed crystals. Each of the novel forms found by means of high-
throughput crystallization was scaled up to multiple milligram and gram levels. Thus the
high-throughput crystallization for solid-form discovery and exploration of large numbers
of parallel crystallization trials led to identification of more polymorphic forms of ritonavir.
Fig. 3. Video micrograph of crystal Form I (left) and Form II (right).
2.2 Nevirapine
Three polymorphs and many solvatomorphs have been reported for nevirapine, a non-
nucleoside reverse transcriptase inhibitor, depending upon the recrystallization method.
Specifically, from a crystal engineering viewpoint, the presence of the amide function CO-
NH in the nevirapine molecule indicates the possibility of alternative modes of self
association, namely via a dimer or a catemer synthon, leading to crystal polymorphism,
whereas interaction with the solvent molecules having complementary donor and acceptor
functions result in formation of solvates, significantly extending the solid state chemistry of
the drug. Form I was prepared by recrystallizing nevirapine from alcohols, ethers, esters or
their mixtures while refluxing from toluene, n-butanol or methyl-isobutyl ketone and
Crystal Forms of Anti-HIV Drugs: Role of Recrystallization
453
subsequent cooling to 0-10 C yielded form II. Form III was prepared by refluxing
nevirapine in chloroform and using dichloromethane as antisolvent to the reaction mixture
(Reguri and Chakka, 2005, 2006). However, many other experiments to investigate the
existence of any other thermodynamically more stable form at room temperature, led to
appearance of different solvatomorphs of nevirapine with varying stoichiometries
depending upon the solvents selected. The first report on preparation of different
solvatomorph by Pereira et al appeared in 2007 (Pereira et al., 2007). Six different solvates of
nevirapine with different morphology were obtained by saturating the solvent systems with
drug at room temperature and cooling in refrigerator. Despite different morphologies the
DSC profiles did not show any relevant differences for raw material and other forms at
crystal fusion peaks. However, a thermal event was observed below the melting
temperature in NEV 3, NEV4 and NEV6 indicating the loss of some solvent molecules. The
presence of solvent is confirmed by TGA and Karl fischer analysis. The authors have
reported the crystal structure of two solvate forms of nevirapine, hemihydrate (NEV3) and
hemiethylacetate (NEV4) (Figure 4).
Fig. 4. Ortep representation of structure of (a) nevirapine hemihydrate and (b) nevirapine
hemiethyl acetate solvate.
Caira et al have also reported five different solvates of nevirapine (Caira et al., 2008). The
initial characterization of all the forms has been performed utilizing DSC and TGA. The
results show that the drug molecule displays significant variability in its modes of self
assembly while accommodating the different solvent molecules. The DSC results show less
thermal stability for toluene solvate while ethylacetate solvate, dichloromethane solvate,
hydrate and 1,4-dioxane solvate were stable. The ethylacetate solvate and dichloromethane
solvate are isostructural and their dimmers are packed in identical fashion, generating
continuous channels parallel to the b-axis that accommodate the solvent molecules (Figure
5). However, in the toluene solvate containing a larger guest molecule, the host dimers are
packed in a different mode, but the guest molecules are again situated in channels.
The effect of series of alcohols on solvate formation capability of nevirapine molecule has
also been illustrated by Caira et al (Caira et al., 2010). The structures of all the solvates were
Recrystallization
454
based on a common isostructural framework comprising centrosymmetric hydrogen-
bonded nevirapine dimers and contain a common channel parallel to the crystal b-axis in
the series which accommodates the various solvent molecules. Thermogravimetric results
yielded a guesthost ratio close to 0.5 for the 1-butanol solvate and a steady decrease in this
ratio from 0.43 to 0.32 for other solvates. This anomalous stoichiometric variation was
resolved following successful X-ray analysis of 1-butanol solvate which revealed that the
length of disordered 1-butanol molecule is proportionate with the channel cavity, resulting
in a stoichiometric association while in other solvates significant disorder for the solvent
molecules was observed which is attributed to their increasing chain lengths being
disproportionate with the channel cavity.
Fig. 5. Inclusion of ethyl acetate molecules within channels in crystals of solvate of
nevirapine.
In the view of tendency of solvate formation of nevirapine molecule, other workers have
also prepared and characterized the various solvates of nevirapine. In one of the studies, the
authors have prepared the solvates by dissolving excess drug in selected solvent systems at
60 C. However, the choice of solvent system was based on the polarity index of the solvents
(Sarkar et al., 2008). Five different crystal forms have been obtained along with an
amorphous form. The crystallization of nevirapine under a variety of crystallization
conditions resulted in a change in the crystal habit of the drug without change in the
internal crystal lattice as suggested by their similar enthalpy of fusion. The mass loss from
the TGA was found to be negligible for all the solvates in comparison to the theoretical mass
loss indicating that solvents used for crystallization formed weak solvates in this study.
However, chadha et al have determined the binding energy of the solvent in the crystal
lattice using differential scanning calorimetry which is found to be higher than the enthalpy
of vapourization of the corresponding solvent for all the solvates except for toluene solvate
indicating that the solvent molecules are tightly bound into the crystal lattice of nevirapine
molecule (except in case of toluene solvate) (Chadha et al., 2010). These authors have also
Crystal Forms of Anti-HIV Drugs: Role of Recrystallization
455
calculated the enthalpy of solvation by determining enthalpy of solution of solvate and drug
in the solvent which is entrapped in the crystal lattice of the solvate using solution
calorimetry technique. The enthalpy of solution when determined in buffer system indicated
that out of the six solvates formed ethanol solvate exhibited the maximum ease for
molecular release of the solvent molecule from the lattice.
2.3 Efavirenz
The abundant hydrogen bonding sites in efavirenz, another non-nucleoside reverse
transcriptase inhibitor, make it a potential candidate to exhibit crystal modifications upon
recrystallization. Driven by this aspect, various authors investigated the solid-state
structures of recrystallized products of efavirenz. The varying recystallization conditions
such as rate of stirring and cooling, antisolvent addition, refluxing/heating, drying under
vacuum, and presence of impurities, along with solvents and/or their mixtures with
varying polarity have yielded different forms of efavirenz. The patent literature till date
reveals 23 different polymorphic forms of efavirenz, one monohydrate and an amorphous
form although there is some ambiguity about the actual number of solid forms (polymorphs
and solvates) of this API (Radesca et al., 1999, 2004; Sharma et al., 2006; Khanduri et al.,
2006; Reddy et al., 2006; Dova, 2008). In these patents, inventors have claimed novel solid
forms of efavirenz based on XRPD and DSC analysis. The patent data shows that Form I is
the most stable form and all the polymorphs revert to this form under some condition or the
other. However, the characterization of these reported forms is not adequate enough to
prove them novel.
Fig. 6. Solid forms of efavirenz, their transformations and some synthons present in the
polymorphs of efavirenz
Recently, two new polymorphc forms structurally characterized by single crystal X-ray
diffraction have been reported by Cuffini et al and Ravikumar et al (Cuffini et al., 2009;
Ravikumar et al., 2009). These two forms do not correspond to the Form I reported in
various patent.
Recrystallization
456
Fig. 7. Efavirenz, form I: (a) A view of the double helical chain viewed down the c-axis with
the three symmetry independent molecules represented in red, green, and blue. The inset
shows the formation of the double helical chains. (b) Hexagonal close packing of helices. (c)
PXRD pattern for form 18g and form I. (d) DSC trace for form I.
The structural information of the stable Form I was first reported by Mahapatra et al
(Mahapatra et al., 2010). The authors obtained this form (Form I) at the interface of an
acetonitrile-water solvent system as well as from methylcyanide-water mixture (Figure 6).
They also reported a solvate of efavirenz which was obtained by dissolving the drug in
cyclohexane at 60 C with stirring and leaving the solution undisturbed for 2 days after
filtration to get rectangular crystals. The DSC analysis of these two forms showed that the
solvate converts to Form I after desolvation. Both the forms have been further characterized
using single crystal X-ray analysis. The efavirenz molecule exhibits a noticeable degree of
conformational disorder in the cyclopropyl group. After exhaustive study it was found that
Form I crystallized in double stranded helices stabilized by N-HO molecules (Figure 7).
However, in the solvate, two molecules of efavirenz form the amide dimer which are
stabilized by C-H interactions to form a bilayer stacked in three dimensions to form
columns. The solvent molecules are lined in these columns and are stabilized by interactions
with CF
3
and cyclopropyl residues (Figure 8).
Crystal Forms of Anti-HIV Drugs: Role of Recrystallization
457
Fig. 8. (a) Crystal packing of efavirenz with the guest cyclohexane molecules represented in
the space filling mode (b) HSM images of II (corresponding temperatures are given in
parentheses)(c) PXRD patterns of forms I and II and the heated sample of form II (d) DSC
plot for form II.
2.4 Lamivudine
Further studies on these categories of drugs have shown the existence of different crystal
forms of lamivudine which is a nucleoside analog reverse transcriptase inhibitor. The solid
state chemistry of this drug is of significant pharmaceutical interest as the drug is reported
to exist in three crystalline forms. The two forms (Form I and II) reported in 1996 were again
studied by Harris et al in 1997 (Jozwiakowski et al., 1996; Harris et al., 1997). Later in 2007, a
new patent showing the existence of another polymorphic form III appeared (Singh et al.,
2007). Michael et al have shown that Form I of lamivudine has been prepared by dissolving
Form II in hot water and then adding an equal volume of methanol to reduce the solubility
of lamivudine. The Form II has been obtained as a result of a synthetic process. The DSC
studies showed heat mediated transformation of Form I to Form II. Besides this, these
authors have also calculated the enthalpy of solution from solubility data and compared the
results with experimently determined enthalpy of solution by solution calorimetry
technique. The enthalpy data revealed that the enthalpy of solution value agrees more
closely in systems of low solubility than in system with high solubility.
Recrystallization
458
Harris et al have recrystallized form I of lamivudine as needles from solutions in water,
methanol or aqueous alcohols while form II as tetragonal bipyramids on slow recrystallization
from dry ethanol and propanol or mixtures of ethanol and less polar organic solvents (Figure
9). Probably, the difference in the polarity of organic solvents led to two different crystalline
forms. The authors have used cross-polarization magic angle spinning (CPMAS) NMR to
differentiate the two forms. The Form II showed a simple spectrum indicating one molecule in
the crystallographic asymmetric unit whereas the spectrum of Form I was found to be
extremely complex due to differences in intermolecular packing environment or
intramolecular geometry/conformation differences. These results were further confirmed by
single crystal X-ray analysis. The Form II was bipyramidal crystals with one molecule in the
asymmetric unit and Form I showed five molecules in the asymmetric unit of crystal lattice.
The crystals of Form III are obtained by subjecting the hot saturated solution of lamivudine
in water to controlled cooling. The DSC and TGA showed this form to be different from
Form I and II. The single crystal X-ray diffraction reveals it to be a hemihydrates with two
molecules of water associated with four molecules of lamivudine in a crystal lattice.
Fig. 9. Scanning electron micrographs of the two forms of Lamivudine
2.5 Stavudine
Stavudine, a thymidine nucleoside reverse transcriptase inhibitor, has been reported to exist
in two anhydrous polymorphic forms and one hydrate. Harte et al and Guruskaya et al have
reported the single crystal of Form I and Form II respectively (Harte et al., 1991 as sited in
Gandhi et al., 2000; Guruskaya et al., 1991 as sited in Gandhi et al., 2000). However, the
production of pure Form I i.,s reported by Gandhi et al who have established the conditions
of recrystallization governing the formation of this thermodynamically most stable form
(Gandhi et al., 2000). The stavudine obtained during the synthesis process is recrystallized
from hot organic solvents as the final step in the synthesis. The cooling of hot isopropanol
solution from 80 to 70 C for over an hour and then to 0-5 C over 1.5 h yielded a mixture of
Form I and II. However, Form I was found to be thermodynamically more stable. Therefore,
some of the crystallization parameters such as rate of cooling or stirring were studied during
recrystallization from isopropanol to selectively obtain Form I. After a lot of
experimentation, it was found that slow cooling of hot isopropanol solution reproducibly
yielded Form I. Kinetically both forms may be present initially but Form II redissolves and
precipitates as Form I with slow cooling.
Crystal Forms of Anti-HIV Drugs: Role of Recrystallization
459
2.6 Cocrystals
Recently, cocrystallization has emerged as an attractive technique to recrystallize molecular
solids that contain two or more distinct chemical components held together by non-covalent
interactions. Anti-HIV agents have also been explored by this approach. The cocrystals of
efavirenz with 4,4-bipyridyl and 1,4-cyclohexanedione prepared by recrystallization of their
grounded mixture from THF and from a mixture of n-heptane and THF respectively have
been reported by Mahapatra et al (Mahapatra et al., 2010). Similarly, the multiple hydrogen
bond donor and acceptor groups of lamivudine and zidovudine have been utilized by Bhatt
et al for designing their cocrystals (Bhatt et al., 2009). These authors have designed
zidovudine cocrystal using retrosynthetic approach where two drug molecules and one
molecule of 2,4,6-triaminopyrimidine are held together by a three point synthon which
forms basic part of their cocrystal structure (Figure 10).
Besides this, a cocrystal hydrate of lamivudine with zidovudine has been reported.
Lamivudine and zidovudine molecules are expected to form synthon II with each other as
shown in figure 11. However, during cocrystallization a hydrated 1:1 cocrystal is formed
and the synthon formed is the extended IIA synthon rather than II. The observed synthon
(IIA) is formed when a molecule of water intervenes in the hydrogen bond pattern of the
synthon II. The cocrystallization without use of water resulted in no cocrystal formation in
this case, perhaps due to the large repulsions between the carbonyl groups in the two API
fragments (Figure 12).
Fig. 10. Three-point synthon I and possibility of this synthon between zidovudine and 2,4,6-
triaminopyrimidine
Fig. 11. Predicted two-point synthon II between lamivudine and zidovudine and observed
synthon IIa in the hydrated co-crystal
Recrystallization
460
Fig. 12. Possible carbonyl - carbonyl repulsion in the putative lamivudine-zidovudine co-
crystal with non-hydrated synthon II
Another cocrystal of lamivudine is designed and prepared with 3,5-dinitrosalicylic acid
based upon synthon containing carboxylic acids and 2-aminopyridines. This cocrystal is an
example of acid base interaction with a very complex hydrogen bond pattern. These authors
also reported cocrystal of lamivudine with 4-quinolinone in a stoichiometry of 1:1. This
cocrystal is stabilized by multiple N-HO and O-H O hydrogen bonds. This cocrystal is
obtained during the screening process and is not based on any synthon theory. Thus the
lamivudine-4-quinolinone cocrystal emphasizes the importance of cocrystal screening to
obtain new cocrystals rather than completely depending on the synthon theory.
Fig. 13. Cocrystals of efavirenz and their preparation
The possible formation of cocrystals of efavirenz with 1,4-cyclohexanedione and 4,4-
bipyridyl was indicated by X-ray diffraction analysis (Mahapatra et al., 2010). The solids
obtained from solvent drop grinding experiment, were subjected to recrystallization in
particular solvent systems (Figure 13). Crystals of efavirenz-1,4-cyclohexanedione were
obtained from heptane-THF solution. These crystals exhibited two symmetry independent
dione molecules in a skewed conformation. One of these two dione molecules makes
bifurcated N-HO/C-HO hydrogen bonds with two symmetry related molecules of
efavirenz, forming a three molecular entity, while the other dione molecule makes linear
chains with bifurcated C-HO hydrogen bonds. These two patterns stacked over one
another to form grill-ribbon structure of the cocrystal (Figure 14).
Crystal Forms of Anti-HIV Drugs: Role of Recrystallization
461
Fig. 14. Cocrystal of efavirenz and 1,4-cyclohexanedione: crystal packing along with the
conformation in the native dione crystal (green).
The cocrystal of efavirenz with 4,4-bipyridyl was obtained by recrystallization from
acetonitrile. Unlike the cocrystal with yclohexanedione in which the major synthon couldnot
be anticipated, in this cocrystal two distinct heterosynthons were observed. While one of the
efavirenz molecules interacted with bipyridyl through cyclic N-HN/C-HO hydrogen
bonds, the other end of pyridine compound made a single point N-HN hydrogen bond
with other molecule of efavirenz resulting in pillered assemblies separated by hydrophobic
cyclopropyl residues (Figure 15).
Recrystallization
462
Fig. 15. Crystal packing.of efavirenz- 4,4-bipyridyl cocrystal
From the above examples it is clear that solvent recrystallization method has been the most
common technique to prepare different crystal forms of a compound. However, with
increasing demands to accelerate the process of crystal form identification, newer and
advanced technologies need to be used as is exemplified by the isolation of three more
polymorphic forms of ritonavir by high-throughput crystallization. In addition,
cocrystallization technique has also been employing solution based crystallization methods
to generate cocrystals of various pharmaceutical compounds. The studies on anti-HIV
agents cited here are thus not simply academic exercises but have practical implications in
preparation and identification of diverse crystalline forms of pharmaceutical material and
can be of considerable benefit in formulation optimization.
3. References
Bauer J., Spanton S., Henry R., Quick J., Dziki W., Porter W., Morris J. Ritonavir: An
extraoridinary example of conformational polymorphism. Pharm. Res., 18, 859-866
(2001).
Bhatt, P.M., Azim, Y., Thakur, T.S., Desiraju, G.R. (2009). Co-crystals of the anti-HIV drugs
lamivudine and zidovudine. Cryst. Growth Des., 9, 951957.
Caira, M.R., Stieger, N., Liebenberg, W., De Villiers, M.M., Samsodien H. (2008). Solvent
inclusion by the anti-HIV drug Nevirapine: X-Ray structures and thermal
decomposition of representative solvates. Cryst. Growth Des., 8, 17-23.
Crystal Forms of Anti-HIV Drugs: Role of Recrystallization
463
Chadha, R., Arora, P., Saini, A., Jain, D.V.S. (2010) Solvated Crystalline Forms of
Nevirapine: Thermoanalytical and Spectroscopic Studies . AAPS PharmSciTech., 11,
3, 1328-1339.
Chemburkar, S. R., Bauer, J., Deming, K., Spiwek, H., Patel, K., Morris, J., Henry, R.,
Spanton, S., Dziki, W., Porter, W., Quick, J., Bauer, P., Donobauer, J., Narayanan, B.
A., Soldani, M., McFarland, D., McFarland K. (2000). Dealing with the impact of
ritonavir polymorphs on the late stages of bulk drug process development. Org.
Process res. deve., 4, 413-417.
Cuffini, S., Howie, R. A., Tiekink, E. R. T., Wardelld, J. L., Wardelle, S. M. S. V. (2009). (S)-6-
Chloro-4-cyclopropylethynyl-4-trifluoromethyl-1H-3,1-benzoxazin-2(4H)-one. Acta
Crystallographica Section E, Cryst., E65, o3170o3171.
Desikan, S., Parsons, Jr. R. L., Davis, W. P., Ward, J. E., Marshall, W. J., Toma, P. H. (2005).
Process development challenges to accommodate a late appearing stable
polymorph: A case study on the polymorphism and crystallization of a fast-track
drug development compound. Org. Process res. deve., 9, 933-942.
Dova, E. (2008). Polymorphic forms of efavirenz. WO 2008/108630 A1.
Gandhi, R.B., Bogardus, D.E., Bugay, D.E., Perrone, R.K., Kalpan, M.A. (2000).
Pharmaceutical relationships of three solid forms of stavudine. Int. J. Pharm., 201,
221-237.
Gurskaya, G.V., Bochkarev, A.V., Ahdanov, A.S., Dyatkina, N.B., Kraevskii, A.A. (1991). X-
ray crystallographic study of 2,3-dideoxy-2,3-didehydrothymidine
conformationally restricted termination substrate of DNA polymerases. Mol. Biol.,
25, 483.
Harris, R. K., Yeung, R. R., Lamont, R. B., Lancaster, R. W., Lynn, S. M., Staniforth, S. E.
(1997). Polymorphism in a novel anti-viral agent: Lamivudine. J. Chem.. Soc.,
Perkin Trans., 2, 2653-2654.
Harte, W.E., Starrett, J.E., Martin, J.C., Mansuri, M.M. (1991). Structure studies of the anti-
HIV agent 2,3-didehydro-2,3-dideoxythymidine (d4T). Biochem. Biophys. Res.
Comm., 175, 298.
Jozwiakowski, M. J., Nguyen, N-A T., Sisco, J. M., Spancake, C. W. (1996). Solubility
behavior of lamivudine crystal forms in recrystallization solvents. J. Pharm. Sci., 85,
193-199.
Khanduri, H. C., Panda, A. K., Kumar, Y. (2006). Processes for the preparation of
polymorphs of efavirenz. WO 2006/030299 A1.
Mahapatra, S., Thakur, T. S., Joseph, S., Varughese, S., Desiraju, G. R. (2010). New Solid State
Forms of the Anti-HIV Drug Efavirenz. Conformational Flexibility and High Z
Issues New Solid State Forms of the Anti-HIV Drug Efavirenz. Conformational
Flexibility and High Z Issues. Cryst. Growth Des., 10, 7, 31913202.
Miller, J. M., Collman, B. M., Greene, L. R., Grant, D. J., Blackburn, A. C. (2005). Identify the
stable polymorph early in drug development process. Pharm. Dev. Technol., 10, 291-
297.
Morissette, S.L., Soukasene, S., Levinson, D., Cima, M.J., Almarsson, O. (2003). Elucidation
of crystal form diversity of the HIV protease inhibitor ritonavir by high-throughput
crystallization. Applied physical sciences, 100, 2181-2184.
Recrystallization
464
Pereira, B. G., Fonte-Boa, F. D., Resende, J.A.L.C., Pinheiro, C.B., Fernandes, N.G., Yoshida,
M.I., Vianna-Soares, C.D. (2007). Pseudopolymorphs and intrinsic dissolution of
Nevirapine. Cryst. Growth Des., 7, 2016-2023.
Radesca, L., Maurin, M., Rabel, S., Moore, J. (1999). Crystalline efavirenz. WO 99/64405.
Radesca, L., Maurin, M., Rabel, S., Moore, J. (2004). Crystalline efavirenz. US 6,673,372 B1.
Ravikumar, K., Sridhar, B. (2009). Molecular and crystal structure of efavirenz, a potent and
specific inhibitor ofHIV-1 reverse transcriptase, and its monohydrate. Mol. Cryst.
Liq. Crystal, 515, 190-198.
Reddy, B.P., Rathnakar, K., Reddy, R.R., Reddy, D.M., Reddy, K.S.C. (2006). Novel
polymorphs of efavirenz. US 2006/0235008.
Reguri, B.R., Chakka, R. (2005). Novel crystalline forms of 11-cyclopropyl-5,11-dihydro-4-
methyl-6H-dipyrido[3,2-b:2,3-e][1,4]diazepin-6-one (Nevirapine). United States
Patent 0059653A1.
Reguri, B.R., Chakka, R. (2006). Crystalline forms of Nevirapine. United States Patent
0183738 A1.
Sarkar, M., Perumal, O.P., Panchagnula, R. (2008). Solid-state characterization of nevirapine.
J. Pharm. Sci., 70, 619-630.
Sharma, R., Bhushan, H.K., Aryan, R.C., Singh, N., Pandya, B., Kumar, Y. (2006).
Polymorphic forms of efavirenz and processes for their preparation. WO
2006/040643 A2.
Singh, G.P., Srivastava, D., Saini, M.B., Upadhyay, P.R. (2007). A novel crystalline form of
lamivudine. WO 2007/119248 A1.
Stieger, N., Liebenberg, W., Wessels, J.C., Samsodien, H. Caira, M. R. (2010). Channel
inclusion of primary alcohols in isostructural solvates of the antiretroviral
nevirapine: an X-ray and thermal analysis study. Stru. Chem., 21, 771-777.